Bump version to 0.4.3

new high level arch abstract
Add 10-node cluster profile to Docker Compose
2026-08-13 15:32:24 -06:00 · 2026-03-07 13:11:42 -08:00 · 2026-03-07 13:01:19 -08:00 · 2026-03-07 12:54:41 -08:00 · 2026-03-07 12:49:07 -08:00 · 2026-03-05 20:18:46 -08:00
83 changed files with 7413 additions and 1284 deletions
@@ -17,3 +17,5 @@ venv/
 .plan.md
 .plan-*.md
 .hypothesis/
+PROGRESS.md
+.coverage
@@ -19,13 +19,9 @@ Turnstone gives LLMs tools — shell, files, search, web, planning — and orche
 - **Cluster dashboard** — real-time view of all nodes and workstreams, workstream creation with node targeting, reverse proxy for server UIs (only the console port needs network access)
 - **Cluster simulator** — test the stack at scale (up to 1000 nodes) without an LLM backend

-```
-External System → Message Queue → Bridge (per node) → Turnstone Server → LLM + Tools
-                                      ↓
-                                 Pub/Sub → Progress Events → External System
-                                      ↓
-                                 turnstone-console → Cluster Dashboard (browser)
-```
+<p align="center">
+  <img src="docs/diagrams/architecture-overview.svg" alt="Turnstone system architecture — data flow from clients through gateways, Redis MQ, cluster nodes, to LLM providers" width="960"/>
+</p>

 ## Quickstart

@@ -111,69 +107,7 @@ All frontends connect to any OpenAI-compatible API (vLLM, NVIDIA NIM/NGC, llama.

 ## Architecture

-```
-turnstone/
-├── core/              # UI-agnostic engine
-│   ├── session.py     # ChatSession — multi-turn loop, tool dispatch, agents
-│   ├── providers/     # LLM provider adapters (OpenAI, Anthropic)
-│   │   ├── _protocol.py   # LLMProvider protocol, ModelCapabilities, StreamChunk
-│   │   ├── _openai.py     # OpenAI-compatible (OpenAI, vLLM, llama.cpp)
-│   │   └── _anthropic.py  # Anthropic Messages API (native streaming, thinking)
-│   ├── tools.py       # Tool definitions (auto-loaded from JSON)
-│   ├── workstream.py  # WorkstreamManager — parallel independent sessions
-│   ├── mcp_client.py  # MCP client manager (external tool servers)
-│   ├── model_registry.py # ModelRegistry — named models, fallback routing, per-workstream selection
-│   ├── config.py      # Unified TOML config (~/.config/turnstone/config.toml)
-│   ├── memory.py      # Persistence facade (delegates to storage/)
-│   ├── storage/       # Pluggable storage backend (SQLite + PostgreSQL)
-│   ├── metrics.py     # Prometheus-compatible metrics collector
-│   ├── healthcheck.py # Backend health monitor + circuit breaker
-│   ├── ratelimit.py   # Per-IP token-bucket rate limiter
-│   ├── edit.py        # File editing (fuzzy match, indentation)
-│   ├── safety.py      # Path validation, sandbox checks
-│   ├── sandbox.py     # Command sandboxing
-│   └── web.py         # Web fetch/search helpers
-├── mq/                # Message queue integration
-│   ├── protocol.py    # Typed message dataclasses (JSON serialization)
-│   ├── broker.py      # Abstract MessageBroker + RedisBroker
-│   ├── bridge.py      # Bridge service (queue ↔ HTTP API, multi-node routing)
-│   └── client.py      # TurnstoneClient — Python API for external systems
-├── console/           # Cluster dashboard
-│   ├── collector.py   # ClusterCollector — aggregates all nodes via Redis + HTTP
-│   ├── server.py      # Dashboard Starlette/ASGI server + SSE
-│   └── static/        # Cluster dashboard web UI
-├── tools/             # Tool schemas (one JSON file per tool)
-├── ui/                # Frontend assets and terminal rendering
-│   └── static/        # Web UI (HTML, CSS, JS)
-├── sim/               # Cluster simulator
-│   ├── cluster.py     # SimCluster — orchestrates N nodes + dispatchers
-│   ├── node.py        # SimNode + SimWorkstream — protocol-compatible node
-│   ├── engine.py      # LLM + tool execution simulation
-│   ├── scenario.py    # 5 workload scenarios (steady, burst, node_failure, …)
-│   ├── metrics.py     # Latency, throughput, utilization collection
-│   └── cli.py         # CLI entry point (turnstone-sim)
-├── cli.py             # Terminal frontend (+ /cluster commands for console)
-├── server.py          # Web frontend (Starlette/ASGI + SSE)
-└── eval.py            # Evaluation and prompt optimization harness
-├── api/               # OpenAPI spec generation (Pydantic v2 models)
-├── sdk/               # Client SDKs (sync + async, Python)
-docs/
-├── architecture.md    # System architecture and threading model
-├── api-reference.md   # Web server API and SSE event reference
-├── sdk.md             # Client SDK reference (Python + TypeScript)
-├── console.md         # Cluster dashboard service (turnstone-console)
-├── docker.md          # Docker Compose deployment and configuration
-├── simulator.md       # Cluster simulator usage and scenarios
-├── tools.md           # Tool schemas, execution pipeline, approval flow
-├── eval.md            # Evaluation harness internals
-└── diagrams/          # UML architecture diagrams (PlantUML sources + PNGs)
-    └── png/           # Pre-rendered diagram images
-deploy/
-├── helm/turnstone/    # Helm chart for Kubernetes
-└── terraform/         # Terraform modules (AWS ECS/Fargate)
-```
-
-### Architecture Diagrams
+### Diagrams

 Detailed UML diagrams are available in [`docs/diagrams/`](docs/diagrams/):

@@ -5,8 +5,8 @@
 #   Default (SQLite):     docker compose up
 #   Production (PG):      DB_BACKEND=postgresql docker compose --profile production up
 #                         (or set DB_BACKEND=postgresql in .env)
+#   10-node cluster:      docker compose --profile cluster up
 #   With simulator:       docker compose --profile sim up
-#   Scale bridges:        docker compose up --scale bridge=3
 # =============================================================================

 name: turnstone
@@ -28,6 +28,7 @@ services:
    image: postgres:17-alpine
    profiles:
      - production
+      - cluster
    environment:
      POSTGRES_DB: turnstone
      POSTGRES_USER: ${POSTGRES_USER:-turnstone}
@@ -213,6 +214,7 @@ services:
        turnstone-channel
        --redis-host=redis
        --redis-port=6379
+        --http-host=0.0.0.0
        $${TURNSTONE_DISCORD_GUILD:+--discord-guild $$TURNSTONE_DISCORD_GUILD}
    environment:
      - TURNSTONE_DISCORD_TOKEN=${TURNSTONE_DISCORD_TOKEN:-}
@@ -221,6 +223,7 @@ services:
      - TURNSTONE_JWT_SECRET=${TURNSTONE_JWT_SECRET:-}
      - TURNSTONE_DB_BACKEND=${DB_BACKEND:-sqlite}
      - TURNSTONE_DB_URL=${DATABASE_URL:-}
+      - TURNSTONE_CHANNEL_ADVERTISE_URL=http://channel:8091
    networks:
      - turnstone-net
    depends_on:
@@ -270,3 +273,219 @@ services:
      redis:
        condition: service_healthy
    restart: "no"
+
+  # ===================================================================
+  # 10-node cluster (profile: cluster)
+  #
+  # Each node is a server + bridge pair.  All share the same PostgreSQL
+  # and Redis instances.  Access via console at :8090.
+  #
+  # Start:  docker compose --profile cluster up
+  # ===================================================================
+
+  # -- cluster servers ------------------------------------------------
+
+  server-1: &cluster-server
+    build: { context: ., dockerfile: Dockerfile }
+    profiles: [cluster]
+    command: &cluster-server-cmd
+      - sh
+      - -c
+      - >-
+        turnstone-server
+        --host 0.0.0.0
+        --port 8080
+        --base-url "$${LLM_BASE_URL}"
+        --api-key "$${OPENAI_API_KEY}"
+        $${MODEL:+--model $$MODEL}
+        $${SKIP_PERMISSIONS:+--skip-permissions}
+    volumes: [turnstone-data:/data]
+    environment: &cluster-server-env
+      LLM_BASE_URL: ${LLM_BASE_URL:-http://host.docker.internal:8000/v1}
+      OPENAI_API_KEY: ${OPENAI_API_KEY:-dummy}
+      TAVILY_API_KEY: ${TAVILY_API_KEY:-}
+      SKIP_PERMISSIONS: ${SKIP_PERMISSIONS:-}
+      TURNSTONE_AUTH_ENABLED: ${TURNSTONE_AUTH_ENABLED:-}
+      TURNSTONE_AUTH_TOKEN: ${TURNSTONE_AUTH_TOKEN:-}
+      TURNSTONE_JWT_SECRET: ${TURNSTONE_JWT_SECRET:-}
+      MODEL: ${MODEL:-}
+      TURNSTONE_DB_BACKEND: ${DB_BACKEND:-postgresql}
+      TURNSTONE_DB_URL: ${DATABASE_URL:-postgresql://${POSTGRES_USER:-turnstone}:${POSTGRES_PASSWORD:?}@postgres:5432/turnstone}
+      TURNSTONE_NODE_ID: node-1
+    extra_hosts: ["host.docker.internal:host-gateway"]
+    networks: [turnstone-net]
+    depends_on:
+      redis: { condition: service_healthy }
+      postgres: { condition: service_healthy }
+    healthcheck:
+      test: ["CMD", "python", "/usr/local/bin/healthcheck.py", "http://127.0.0.1:8080/health"]
+      interval: 10s
+      timeout: 5s
+      retries: 3
+      start_period: 15s
+    deploy:
+      resources:
+        limits: { memory: 384M, cpus: '0.5' }
+    restart: unless-stopped
+
+  server-2:
+    <<: *cluster-server
+    environment: { <<: *cluster-server-env, TURNSTONE_NODE_ID: node-2 }
+  server-3:
+    <<: *cluster-server
+    environment: { <<: *cluster-server-env, TURNSTONE_NODE_ID: node-3 }
+  server-4:
+    <<: *cluster-server
+    environment: { <<: *cluster-server-env, TURNSTONE_NODE_ID: node-4 }
+  server-5:
+    <<: *cluster-server
+    environment: { <<: *cluster-server-env, TURNSTONE_NODE_ID: node-5 }
+  server-6:
+    <<: *cluster-server
+    environment: { <<: *cluster-server-env, TURNSTONE_NODE_ID: node-6 }
+  server-7:
+    <<: *cluster-server
+    environment: { <<: *cluster-server-env, TURNSTONE_NODE_ID: node-7 }
+  server-8:
+    <<: *cluster-server
+    environment: { <<: *cluster-server-env, TURNSTONE_NODE_ID: node-8 }
+  server-9:
+    <<: *cluster-server
+    environment: { <<: *cluster-server-env, TURNSTONE_NODE_ID: node-9 }
+  server-10:
+    <<: *cluster-server
+    environment: { <<: *cluster-server-env, TURNSTONE_NODE_ID: node-10 }
+
+  # -- cluster bridges ------------------------------------------------
+
+  bridge-1: &cluster-bridge
+    build: { context: ., dockerfile: Dockerfile }
+    profiles: [cluster]
+    command:
+      - turnstone-bridge
+      - --server-url=http://server-1:8080
+      - --redis-host=redis
+      - --redis-port=6379
+      - --heartbeat-ttl=${HEARTBEAT_TTL:-60}
+      - --approval-timeout=${APPROVAL_TIMEOUT:-3600}
+    environment: &cluster-bridge-env
+      REDIS_PASSWORD: ${REDIS_PASSWORD:-}
+      TURNSTONE_AUTH_TOKEN: ${TURNSTONE_AUTH_TOKEN:-}
+      TURNSTONE_JWT_SECRET: ${TURNSTONE_JWT_SECRET:-}
+    networks: [turnstone-net]
+    depends_on:
+      server-1: { condition: service_healthy }
+      redis: { condition: service_healthy }
+    deploy:
+      resources:
+        limits: { memory: 256M, cpus: '0.25' }
+    restart: unless-stopped
+
+  bridge-2:
+    <<: *cluster-bridge
+    command:
+      - turnstone-bridge
+      - --server-url=http://server-2:8080
+      - --redis-host=redis
+      - --redis-port=6379
+      - --heartbeat-ttl=${HEARTBEAT_TTL:-60}
+      - --approval-timeout=${APPROVAL_TIMEOUT:-3600}
+    depends_on:
+      server-2: { condition: service_healthy }
+      redis: { condition: service_healthy }
+  bridge-3:
+    <<: *cluster-bridge
+    command:
+      - turnstone-bridge
+      - --server-url=http://server-3:8080
+      - --redis-host=redis
+      - --redis-port=6379
+      - --heartbeat-ttl=${HEARTBEAT_TTL:-60}
+      - --approval-timeout=${APPROVAL_TIMEOUT:-3600}
+    depends_on:
+      server-3: { condition: service_healthy }
+      redis: { condition: service_healthy }
+  bridge-4:
+    <<: *cluster-bridge
+    command:
+      - turnstone-bridge
+      - --server-url=http://server-4:8080
+      - --redis-host=redis
+      - --redis-port=6379
+      - --heartbeat-ttl=${HEARTBEAT_TTL:-60}
+      - --approval-timeout=${APPROVAL_TIMEOUT:-3600}
+    depends_on:
+      server-4: { condition: service_healthy }
+      redis: { condition: service_healthy }
+  bridge-5:
+    <<: *cluster-bridge
+    command:
+      - turnstone-bridge
+      - --server-url=http://server-5:8080
+      - --redis-host=redis
+      - --redis-port=6379
+      - --heartbeat-ttl=${HEARTBEAT_TTL:-60}
+      - --approval-timeout=${APPROVAL_TIMEOUT:-3600}
+    depends_on:
+      server-5: { condition: service_healthy }
+      redis: { condition: service_healthy }
+  bridge-6:
+    <<: *cluster-bridge
+    command:
+      - turnstone-bridge
+      - --server-url=http://server-6:8080
+      - --redis-host=redis
+      - --redis-port=6379
+      - --heartbeat-ttl=${HEARTBEAT_TTL:-60}
+      - --approval-timeout=${APPROVAL_TIMEOUT:-3600}
+    depends_on:
+      server-6: { condition: service_healthy }
+      redis: { condition: service_healthy }
+  bridge-7:
+    <<: *cluster-bridge
+    command:
+      - turnstone-bridge
+      - --server-url=http://server-7:8080
+      - --redis-host=redis
+      - --redis-port=6379
+      - --heartbeat-ttl=${HEARTBEAT_TTL:-60}
+      - --approval-timeout=${APPROVAL_TIMEOUT:-3600}
+    depends_on:
+      server-7: { condition: service_healthy }
+      redis: { condition: service_healthy }
+  bridge-8:
+    <<: *cluster-bridge
+    command:
+      - turnstone-bridge
+      - --server-url=http://server-8:8080
+      - --redis-host=redis
+      - --redis-port=6379
+      - --heartbeat-ttl=${HEARTBEAT_TTL:-60}
+      - --approval-timeout=${APPROVAL_TIMEOUT:-3600}
+    depends_on:
+      server-8: { condition: service_healthy }
+      redis: { condition: service_healthy }
+  bridge-9:
+    <<: *cluster-bridge
+    command:
+      - turnstone-bridge
+      - --server-url=http://server-9:8080
+      - --redis-host=redis
+      - --redis-port=6379
+      - --heartbeat-ttl=${HEARTBEAT_TTL:-60}
+      - --approval-timeout=${APPROVAL_TIMEOUT:-3600}
+    depends_on:
+      server-9: { condition: service_healthy }
+      redis: { condition: service_healthy }
+  bridge-10:
+    <<: *cluster-bridge
+    command:
+      - turnstone-bridge
+      - --server-url=http://server-10:8080
+      - --redis-host=redis
+      - --redis-port=6379
+      - --heartbeat-ttl=${HEARTBEAT_TTL:-60}
+      - --approval-timeout=${APPROVAL_TIMEOUT:-3600}
+    depends_on:
+      server-10: { condition: service_healthy }
+      redis: { condition: service_healthy }
@@ -515,8 +515,8 @@ Returns a list of all active workstreams.
 ```json
 {
  "workstreams": [
-    {"id": "abc123", "name": "default", "state": "idle", "session_id": "a1b2c3d4e5f6"},
-    {"id": "def456", "name": "hacker-news", "state": "thinking", "session_id": "c5d6e7f8a9b0"}
+    {"id": "abc123", "name": "default", "state": "idle"},
+    {"id": "def456", "name": "hacker-news", "state": "thinking"}
  ]
 }
 ```
@@ -528,22 +528,21 @@ Each workstream object:
 | `id`         | string      | Unique workstream routing identifier                   |
 | `name`       | string      | Display name (alias if set, otherwise `ws-xxxx`)       |
 | `state`      | string      | Current state (see state values above)                 |
-| `session_id` | string/null | Session ID of the workstream's `ChatSession`, used for deduplication against `/v1/api/sessions` |

 ---

-### `GET /v1/api/sessions`
+### `GET /v1/api/workstreams/saved`

-Returns a list of saved sessions from the database, ordered by most recently
+Returns a list of saved workstreams from the database, ordered by most recently
 updated.

 **Response:**

 ```json
 {
-  "sessions": [
+  "workstreams": [
    {
-      "session_id": "a1b2c3d4e5f6",
+      "ws_id": "a1b2c3d4e5f6",
      "alias": "refactor",
      "title": "JWT Authentication Refactor",
      "created": "2026-03-01 10:00:00",
@@ -554,18 +553,16 @@ updated.
 }
 ```

-Each session object:
+Each saved workstream object:

 | Field           | Type        | Description                                |
 |-----------------|-------------|--------------------------------------------|
-| `session_id`    | string      | Unique 32-char hex UUID session identifier |
+| `ws_id`         | string      | Unique workstream identifier               |
 | `alias`         | string/null | User-assigned short name                   |
 | `title`         | string/null | LLM-generated title                        |
-| `created`       | string      | ISO timestamp of session creation          |
+| `created`       | string      | ISO timestamp of workstream creation       |
 | `updated`       | string      | ISO timestamp of last message              |
-| `message_count` | int         | Number of messages in the session          |
-| `node_id`       | string/null | Server node that created the session       |
-| `ws_id`         | string/null | Workstream the session belongs to          |
+| `message_count` | int         | Number of messages in the workstream       |

 ---

@@ -721,7 +718,7 @@ All fields are optional. The body can be empty or an empty JSON object.
 | `name`           | string | auto    | Workstream display name                                        |
 | `model`          | string | default | Model alias from the registry (`[models.*]`)                   |
 | `auto_approve`   | bool   | false   | Auto-approve all tool calls for this workstream                |
-| `resume_session` | string | ""      | Session ID to resume atomically during creation (empty = fresh)|
+| `resume_ws`      | string | ""      | Workstream ID to resume atomically during creation (empty = fresh)|

 **Response (success):**

@@ -75,6 +75,7 @@ turnstone/
    client.py         TurnstoneClient library + TurnResult for MQ-based access
  console/
    collector.py      ClusterCollector — aggregates state from all nodes via Redis + HTTP
+    scheduler.py      TaskScheduler — background cron/at scheduler, dispatches via MQ
    server.py         Cluster dashboard HTTP server + SSE + CLI entry point
    static/           Cluster dashboard web UI (page-specific HTML, CSS, JS)
  channels/
@@ -471,7 +472,7 @@ independently, then returns the final content as the tool result.

 - **task**: uses `self._task_tools` (`TASK_AGENT_TOOLS` + MCP tools)
 - **plan**: uses `self._agent_tools` (`AGENT_TOOLS` + MCP tools). Writes output
-  to `.plan-<session_id>.md` — unique per `ChatSession` so concurrent workstreams
+  to `.plan-<ws_id>.md` — unique per `ChatSession` so concurrent workstreams
  don't collide. On repeat invocations the prior `plan` tool call and its result
  are forwarded from `self.messages` so the agent refines the existing plan rather
  than starting over. Planning instructions are injected as a developer message
@@ -687,8 +688,11 @@ memories
  created  TEXT NOT NULL
  updated  TEXT NOT NULL

-sessions
-  session_id  TEXT PRIMARY KEY
+workstreams
+  ws_id       TEXT PRIMARY KEY
+  node_id     TEXT NOT NULL
+  name        TEXT NOT NULL
+  state       TEXT NOT NULL DEFAULT 'idle'
  alias       TEXT UNIQUE            -- user-assigned short name (nullable)
  title       TEXT                   -- LLM-generated title (nullable)
  created     TEXT NOT NULL
@@ -696,7 +700,7 @@ sessions

 conversations
  id            INTEGER PRIMARY KEY AUTOINCREMENT
-  session_id    TEXT NOT NULL
+  ws_id         TEXT NOT NULL
  timestamp     TEXT NOT NULL
  role          TEXT NOT NULL        -- user | assistant | tool_call | tool_result
  content       TEXT
@@ -705,8 +709,8 @@ conversations
  tool_call_id  TEXT                 -- links tool_call ↔ tool_result for resume
  provider_data TEXT                 -- raw provider content (e.g. Anthropic encrypted)

-session_config
-  session_id  TEXT NOT NULL          -- composite PK with key
+workstream_config
+  ws_id       TEXT NOT NULL          -- composite PK with key
  key         TEXT NOT NULL
  value       TEXT

@@ -721,22 +725,20 @@ and are the single source of truth for both backends and Alembic migrations.

 | Method | Purpose |
 |--------|---------|
-| `register_session(session_id, title, node_id, ws_id)` | Create a sessions row (no-op if exists) |
-| `save_message(session_id, role, content, ...)` | Log a message to conversations |
-| `load_session_messages(session_id)` | Reconstruct OpenAI message format from DB rows |
-| `list_sessions(limit)` | List sessions with >=1 message, ordered by updated DESC |
-| `delete_session(session_id)` | Delete session and all its messages |
-| `prune_sessions(retention_days)` | Remove empty sessions and old unnamed sessions |
-| `resolve_session(alias_or_id)` | Resolve alias, exact id, or id prefix to full session_id |
-| `save_session_config(session_id, config)` | Persist session configuration key/value pairs |
-| `load_session_config(session_id)` | Retrieve session configuration |
-| `set_session_alias(session_id, alias)` | Set user-friendly alias (returns False if taken) |
-| `get_session_name(session_id)` | Return alias if set, else title, else None |
-| `update_session_title(session_id, title)` | Set/update LLM-generated title |
 | `register_workstream(ws_id, node_id, name, state)` | Create a workstreams row (no-op if exists) |
+| `save_message(ws_id, role, content, ...)` | Log a message to conversations |
+| `load_messages(ws_id)` | Reconstruct OpenAI message format from DB rows |
+| `list_workstreams_with_history(limit)` | List workstreams with >=1 message, ordered by updated DESC |
+| `delete_workstream(ws_id)` | Delete workstream and cascade conversations + config |
+| `prune_workstreams(retention_days)` | Remove empty workstreams and old unnamed workstreams |
+| `resolve_workstream(alias_or_id)` | Resolve alias, exact id, or id prefix to full ws_id |
+| `save_workstream_config(ws_id, config)` | Persist workstream configuration key/value pairs |
+| `load_workstream_config(ws_id)` | Retrieve workstream configuration |
+| `set_workstream_alias(ws_id, alias)` | Set user-friendly alias (returns False if taken) |
+| `get_workstream_display_name(ws_id)` | Return alias if set, else title, else None |
+| `update_workstream_title(ws_id, title)` | Set/update LLM-generated title |
 | `update_workstream_state(ws_id, state)` | Update workstream state and bump timestamp |
 | `update_workstream_name(ws_id, name)` | Update workstream display name |
-| `delete_workstream(ws_id)` | Delete a workstream row |
 | `list_workstreams(node_id, limit)` | List workstreams, optionally by node |
 | `kv_get(key)` / `kv_set(key, value)` / `kv_delete(key)` | Generic key-value store (backs memories table) |
 | `kv_list()` / `kv_search(query)` | List or search key-value pairs |
@@ -756,59 +758,59 @@ pool_size = 5                       # PostgreSQL connection pool size

 Environment variables: `TURNSTONE_DB_BACKEND`, `TURNSTONE_DB_URL`, `TURNSTONE_DB_PATH`.

-### Session Persistence and Resume
+### Persistence and Resume

-Each `ChatSession` generates a full 32-char hex UUID `_session_id` on creation
-and registers it in the `sessions` table with the server's `node_id` and the
-owning `ws_id`. Messages are saved to `conversations` as they happen via
-`save_message()`. Workstreams are persisted to the `workstreams` table on
-creation, with state changes tracked via `update_workstream_state()`.
+`ws_id` is the sole persistent identity for both routing and conversation
+history. There is no separate `session_id` — the `workstreams` table holds
+alias, title, and state alongside the routing fields (`node_id`, `name`).
+Messages are saved to `conversations` (keyed by `ws_id`) as they happen
+via `save_message()`. Workstream state changes are tracked via
+`update_workstream_state()`.

 **Auto-titling:** After the first complete exchange (user message + assistant
 response), a background thread calls the LLM with a title-generation prompt
 (`reasoning_effort: "low"`, `max_completion_tokens: 200`). The generated
-title (3-8 words) is stored in `sessions.title`.
+title (3-8 words) is stored in `workstreams.title`.

-**Resume flow:** `ChatSession.resume_session(session_id)` calls
-`load_session_messages()` which reconstructs the OpenAI message format from
-database rows:
+**Resume flow:** `ChatSession.resume(ws_id)` calls `load_messages()` which
+reconstructs the OpenAI message format from database rows:

 - `user` and `assistant` rows map directly
 - Consecutive `tool_call` rows are grouped into one assistant message's
  `tool_calls` array, paired with subsequent `tool_result` rows via
  `tool_call_id` (or positional matching for legacy data)
- **Interrupted session repair:** If the last assistant message has
-  `tool_calls` but fewer tool results than expected (session was
+- **Interrupted conversation repair:** If the last assistant message has
+  `tool_calls` but fewer tool results than expected (conversation was
  interrupted mid-execution), the incomplete turn is stripped so the
  LLM can re-generate cleanly
- The session adopts the old `_session_id`, so new messages continue in
-  the same session
+- The `ChatSession` adopts the resumed `_ws_id`, so new messages continue
+  in the same workstream

 **Config persistence:** LLM-affecting parameters (`temperature`,
 `reasoning_effort`, `max_tokens`, `instructions`, `creative_mode`) are
-persisted to the `session_config` table on creation and whenever changed
-via slash commands. `resume_session()` restores these values so resumed
-sessions behave identically to the original.
+persisted to the `workstream_config` table on creation and whenever changed
+via slash commands. `resume()` restores these values so resumed workstreams
+behave identically to the original.

 **`/clear` vs `/new`:** `/clear` wipes in-memory context but preserves
-messages in the database for future resume. `/new` starts a fresh session
-(new `_session_id`), leaving the old session resumable.
+messages in the database for future resume. `/new` starts a fresh workstream
+(new `_ws_id`), leaving the old workstream resumable.

-**Resolution:** `resolve_session()` accepts aliases, exact session IDs, or
-session ID prefixes, enabling `turnstone --resume refactor` or `/resume abc12`.
+**Resolution:** `resolve_workstream()` accepts aliases, exact workstream IDs,
+or ID prefixes, enabling `turnstone --resume refactor` or `/resume abc12`.

-**Session listing:** `list_sessions()` only returns sessions that have at
-least one saved message (`WHERE EXISTS` on `conversations`). Sessions
-registered but never used (e.g., from process startup) are invisible until
-a message is sent.
+**Workstream listing:** `list_workstreams_with_history()` only returns
+workstreams that have at least one saved message (`WHERE EXISTS` on
+`conversations`). Workstreams registered but never used (e.g., from process
+startup) are invisible until a message is sent.

-**Session pruning:** `prune_sessions(retention_days, log_fn)` runs once at
-startup (CLI and server). It removes:
- Sessions with no messages (orphaned registrations)
- Unnamed sessions (`alias IS NULL`) older than `retention_days` days (default 90)
+**Workstream pruning:** `prune_workstreams(retention_days, log_fn)` runs once
+at startup (CLI and server). It removes:
+- Workstreams with no messages (orphaned registrations)
+- Unnamed workstreams (`alias IS NULL`) older than `retention_days` days (default 90)

-Named (aliased) sessions are never age-pruned. Configure with
-`--session-retention-days N` (0 = disable age pruning).
+Named (aliased) workstreams are never age-pruned. Configure with
+`--retention-days N` (0 = disable age pruning).

 ---

@@ -946,7 +948,7 @@ Three hierarchical scopes control endpoint access:

 | Scope | Grants | Endpoints |
 |-------|--------|-----------|
-| `read` | SSE streams, workstream listing, sessions | GET endpoints |
+| `read` | SSE streams, workstream listing, history | GET endpoints |
 | `write` | `read` + send, command, workstream create/close | POST to `/api/send`, `/api/command`, etc. |
 | `approve` | `write` + tool approval, admin operations | POST to `/api/approve`, `/api/admin/*` |

@@ -1238,7 +1240,7 @@ typed event dataclasses.

 **Two client pairs** (sync + async):

- `TurnstoneServer` / `AsyncTurnstoneServer` — server API (workstreams, chat, streaming, sessions)
+- `TurnstoneServer` / `AsyncTurnstoneServer` — server API (workstreams, chat, streaming)
 - `TurnstoneConsole` / `AsyncTurnstoneConsole` — console API (cluster overview, nodes, workstreams)

 **Design**: async-first with thin sync wrappers. `_BaseClient` provides httpx
@@ -1278,12 +1280,26 @@ The `ChannelRouter` manages bidirectional routing: it maps platform
 channel/thread IDs to turnstone workstream IDs, handles workstream
 creation and stale-route recovery, and resolves platform users to
 turnstone identities via the `channel_users` table. When an evicted
-workstream is reactivated, the router uses atomic session resume via the
-`resume_session` field on `CreateWorkstreamMessage` — the server resumes
-the old session during workstream creation in a single HTTP request,
-eliminating ordering fragility. The bridge emits a `SessionResumedEvent`
-to confirm success.
+workstream is reactivated, the router uses atomic resume via the
+`resume_ws` field on `CreateWorkstreamMessage` — the server resumes
+the old workstream's conversation during creation in a single HTTP
+request, eliminating ordering fragility. The bridge emits a
+`WorkstreamResumedEvent` to confirm success.

 Discord ships as the first adapter. See [channels.md](channels.md) for
 setup instructions, configuration reference, and the adapter development
 guide.
+
+### Notification Subsystem
+
+The `notify` tool enables the LLM to send notifications to users or
+channels without going through MQ. The server calls the channel gateway
+directly over HTTP for lower latency: `_exec_notify()` queries the
+`services` database table for healthy channel gateways (heartbeat within
+120 seconds), authenticates with a service JWT (`aud: turnstone-channel`),
+and POSTs to `POST /v1/api/notify` on the first healthy gateway. The
+gateway validates the JWT, resolves the target (username lookup via
+`channel_users` or direct `channel_type`+`channel_id`), and delegates to
+the appropriate `ChannelAdapter.send()`. Delivery retries up to 3 times
+with backoff, re-querying the service registry on each attempt. See
+[Notification Flow diagram](diagrams/png/17-notify-flow.png).
@@ -136,11 +136,11 @@ An admin can also force-link or unlink users via the console admin panel
  1.5 seconds.
 - If the workstream is evicted for capacity, the next message in the
  thread auto-creates a new workstream and atomically resumes the
-  previous session via the `resume_session` field on
-  `CreateWorkstreamMessage`. The server resumes the session during
-  workstream creation (same HTTP request), and the bridge emits a
-  `SessionResumedEvent` back to the channel. The thread receives a
-  *"Session resumed: {name} ({count} messages restored)"* confirmation.
+  previous workstream via the `resume_ws` field on
+  `CreateWorkstreamMessage`. The server resumes the workstream during
+  creation (same HTTP request), and the bridge emits a
+  `WorkstreamResumedEvent` back to the channel. The thread receives a
+  *"Resumed: {name} ({count} messages restored)"* confirmation.

 ### Slash Commands

@@ -198,6 +198,9 @@ Plan review requests are displayed as a blue embed with:
 | `--redis-db` | — | `0` | Redis DB number |
 | `--model` | — | server default | Default model for new workstreams |
 | `--auto-approve` | — | `false` | Auto-approve ALL tool calls (skips approval buttons entirely) |
+| `--http-host` | — | `127.0.0.1` | HTTP server bind address for notify endpoint |
+| `--http-port` | `TURNSTONE_CHANNEL_PORT` | `8091` | HTTP server port |
+| `--auth-token` | `TURNSTONE_CHANNEL_AUTH_TOKEN` | — | Static auth token for `/v1/api/notify` (alternative to JWT) |
 | `--log-level` | `TURNSTONE_LOG_LEVEL` | `INFO` | Log level |
 | `--log-format` | `TURNSTONE_LOG_FORMAT` | `auto` | Log format (`auto`/`json`/`text`) |

@@ -229,19 +232,90 @@ See [Security: Database Schema](security.md#database-schema) for the
 3. **Eviction** — the server evicts an idle workstream for capacity. The
   route is preserved and the thread stays open.
 4. **Reactivation** — the next message in the thread detects the stale
-   route (no MQ owner), looks up the old session via
-   `get_session_id_by_ws()`, and creates a new workstream with
-   `resume_session` set atomically on the `CreateWorkstreamMessage`. The
-   server resumes the session during creation (no separate command
-   needed). The bridge emits a `SessionResumedEvent` to the channel, and
-   the thread displays *"Session resumed: {name} ({count} messages
-   restored)"*. If the old session was pruned, the workstream starts
-   fresh with no error.
+   route (no MQ owner) and creates a new workstream with the old `ws_id`
+   as `resume_ws` on the `CreateWorkstreamMessage`. The server resumes
+   the workstream during creation (no separate command or reverse lookup
+   needed). The bridge emits a `WorkstreamResumedEvent` to the channel, and
+   the thread displays *"Resumed: {name} ({count} messages restored)"*.
+   If the old workstream was pruned, a fresh one starts with no error.
 5. **Close** — `/close` command closes the workstream via MQ, deletes the
   route, unsubscribes from events, and archives the Discord thread.

 ---

+## Notifications
+
+> See also: [Notification Flow diagram](diagrams/png/17-notify-flow.png)
+
+The `notify` tool allows the LLM to proactively send notifications to
+users or channels on external platforms. This is useful for alerting
+people about task completion, errors, or important updates without
+waiting for them to check in.
+
+### Targeting
+
+Two modes:
+
+- **Username** — provide a turnstone `username`. The gateway resolves
+  it via the `channel_users` table and sends to all linked channels
+  (e.g. Discord + future Slack).
+- **Direct** — provide `channel_type` + `channel_id` to target a
+  specific platform channel or user DM.
+
+### Delivery Flow
+
+Notifications bypass MQ for lower latency. The server calls the channel
+gateway directly over HTTP:
+
+1. The LLM calls the `notify` tool with a message and target
+2. `_exec_notify()` queries the `services` table for healthy channel
+   gateways (heartbeat within the last 120 seconds)
+3. The server mints a service JWT (`aud: turnstone-channel`) via
+   `ServiceTokenManager` and POSTs to the first healthy gateway
+4. The gateway validates the JWT, resolves the target, and calls
+   `adapter.send()` on the appropriate platform adapter
+5. On failure, the server tries the next gateway. If all fail, it
+   retries up to 2 more times (delays: 1s, 3s), re-querying the
+   service registry on each attempt
+
+### Service Registry
+
+The channel gateway registers itself in the `services` database table
+on startup and sends a heartbeat every 30 seconds. On shutdown it
+deregisters. Services are considered stale after 120 seconds (4 missed
+heartbeats) and are excluded from `list_services()` queries.
+
+The `services` table schema:
+
+| Column | Description |
+|--------|-------------|
+| `service_type` | Service category (e.g. `"channel"`) |
+| `service_id` | Unique instance ID (`channel-<hostname>-<random>`) |
+| `url` | HTTP base URL for the service |
+| `last_heartbeat` | ISO 8601 timestamp of last heartbeat |
+| `created` | ISO 8601 timestamp of initial registration |
+
+### Security
+
+- **Authentication** — the gateway's `POST /v1/api/notify` endpoint
+  requires authentication. Configure either `TURNSTONE_JWT_SECRET`
+  (the server mints JWTs with `aud: turnstone-channel` automatically)
+  or a static token via `--auth-token`. If neither is set, the
+  gateway fails closed and rejects all requests with 401. Server JWTs
+  (`aud: turnstone-server`) are rejected.
+- **Rate limit** — maximum 5 notifications per turn. The counter only
+  increments on successful delivery, so failures don't consume the
+  budget.
+- **SSRF protection** — only `http://` and `https://` service URLs
+  are allowed. Other schemes are silently skipped.
+- **Mention sanitization** — `discord.utils.escape_mentions()` is
+  applied before sending, preventing `@everyone` / `@here` abuse.
+- **Error redaction** — generic error messages are returned to the
+  LLM. Internal details (service IDs, URLs, exception messages) are
+  logged server-side only.
+
+---
+
 ## Adding New Adapters

 The `ChannelAdapter` protocol defines the interface any platform adapter
@@ -423,6 +423,150 @@ to create the initial admin user and receive a JWT in one step. See

 ---

+## Scheduled Tasks
+
+The console includes a background **TaskScheduler** daemon that creates workstreams on a timed basis via the MQ broker. It supports cron-based recurring schedules and one-shot `at` schedules.
+
+### Architecture
+
+The scheduler runs as a daemon thread inside the console process. Every `check_interval` seconds (default 15) it:
+
+1. Acquires a distributed lock via Redis `SET NX EX` (prevents duplicate dispatch in multi-console deployments)
+2. Queries the storage backend for tasks whose `next_run <= now` and `enabled = true`
+3. Dispatches each due task as one or more `CreateWorkstreamMessage` via MQ
+4. Updates `last_run` and computes the next `next_run` (or disables one-shot `at` tasks)
+5. Releases the lock via Lua script (safe conditional delete)
+
+Run history is automatically pruned (runs older than 90 days) approximately once per hour.
+
+### Schedule Types
+
+| Type | Field | Behavior |
+|------|-------|----------|
+| `cron` | `cron_expr` | Recurring schedule using standard 5-field cron syntax. Requires `croniter`. |
+| `at` | `at_time` | One-shot: fires once at the given ISO 8601 timestamp (must include timezone), then auto-disables. |
+
+### Target Modes
+
+| Mode | Behavior |
+|------|----------|
+| `auto` | Picks the reachable node with the most available capacity |
+| `pool` | Pushes to the shared inbound queue (any bridge picks it up) |
+| `all` | Fan-out to all reachable nodes (capped at `max_fan_out`, default 20) |
+| `<node_id>` | Targets a specific node by ID |
+
+### Configuration
+
+| Parameter | Default | Description |
+|-----------|---------|-------------|
+| `check_interval` | `15.0` | Seconds between scheduler ticks |
+| `lock_ttl` | `60` | Distributed lock TTL in seconds |
+| `max_fan_out` | `20` | Maximum nodes for `all` target mode |
+
+Dependency: `croniter` (installed with turnstone).
+
+### Schedule API
+
+All schedule endpoints require `approve` scope. Maximum 200 schedules.
+
+#### `GET /v1/api/admin/schedules`
+
+List all scheduled tasks.
+
+```json
+{
+  "schedules": [
+    {
+      "task_id": "a1b2c3d4",
+      "name": "nightly-checks",
+      "description": "Run nightly health checks",
+      "schedule_type": "cron",
+      "cron_expr": "0 2 * * *",
+      "at_time": "",
+      "target_mode": "auto",
+      "model": "",
+      "initial_message": "Run the nightly health check suite.",
+      "auto_approve": false,
+      "auto_approve_tools": [],
+      "enabled": true,
+      "created_by": "u_admin",
+      "last_run": "2026-03-05T02:00:00Z",
+      "next_run": "2026-03-06T02:00:00Z",
+      "created": "2026-03-01T12:00:00Z",
+      "updated": "2026-03-05T02:00:01Z"
+    }
+  ]
+}
+```
+
+#### `POST /v1/api/admin/schedules`
+
+Create a scheduled task.
+
+Request:
+
+```json
+{
+  "name": "nightly-checks",
+  "description": "Run nightly health checks",
+  "schedule_type": "cron",
+  "cron_expr": "0 2 * * *",
+  "target_mode": "auto",
+  "initial_message": "Run the nightly health check suite.",
+  "auto_approve": false,
+  "enabled": true
+}
+```
+
+Required fields: `name`, `schedule_type`, `initial_message`. For `cron` schedules provide `cron_expr`; for `at` schedules provide `at_time` (ISO 8601 with timezone, must be in the future).
+
+Response: `ScheduleInfo` (same shape as list items above). Returns `400` for invalid cron syntax, naive timestamps, or past `at_time`. Returns `409` if the 200-schedule cap is reached.
+
+#### `GET /v1/api/admin/schedules/{task_id}`
+
+Get a single scheduled task. Returns `ScheduleInfo` or `404`.
+
+#### `PUT /v1/api/admin/schedules/{task_id}`
+
+Partial update — only include fields to change. If `schedule_type`, `cron_expr`, or `at_time` change, `next_run` is recomputed automatically.
+
+```json
+{
+  "enabled": false
+}
+```
+
+Response: updated `ScheduleInfo`. Returns `400` for validation errors, `404` if not found.
+
+#### `DELETE /v1/api/admin/schedules/{task_id}`
+
+Delete a scheduled task and all its run history. Returns `{"status": "ok"}` or `404`.
+
+#### `GET /v1/api/admin/schedules/{task_id}/runs?limit=50`
+
+List execution history for a task (most recent first). `limit` defaults to 50, max 200.
+
+```json
+{
+  "runs": [
+    {
+      "run_id": "r_abc123",
+      "task_id": "a1b2c3d4",
+      "node_id": "db-west-04",
+      "ws_id": "ws_xyz",
+      "correlation_id": "corr_789",
+      "started": "2026-03-05T02:00:00Z",
+      "status": "dispatched",
+      "error": ""
+    }
+  ]
+}
+```
+
+Status is `dispatched` on success or `failed` with an `error` message (e.g. no reachable nodes). Failed runs do not advance `next_run`.
+
+---
+
 ## CLI Commands

 The `/cluster` command in the turnstone CLI queries the console's HTTP API. Requires `--console-url` or `[console] url` in config.toml.
@@ -118,7 +118,7 @@ class "ChatSession" as ChatSession {
    - ui: SessionUI
    - messages: list[dict]
    - _msg_tokens: list[int]
-    - _session_id: str
+    - _ws_id: str
    - _mcp_client: MCPClientManager | None
    - _registry: ModelRegistry | None
    + model_alias: str | None {property}
@@ -130,7 +130,7 @@ class "ChatSession" as ChatSession {
    --
    + send(user_input: str)
    + handle_command(command: str)
-    + resume_session(session_id: str)
+    + resume(ws_id: str)
    - _save_config()
    - _stream_response(stream) → dict
    - _create_stream_with_retry(msgs) → Stream (+ fallback)
@@ -18,7 +18,7 @@ User -> CS : send(user_input)
 activate CS

 CS -> CS : messages.append({role: "user", content: input})
-CS -> DB : save_message(session_id, "user", input)
+CS -> DB : save_message(ws_id, "user", input)

 == LLM Call Loop ==

@@ -65,8 +65,8 @@ group loop [while tool_calls present]

    CS -> CS : _update_token_table()\ncalibrate chars_per_token ratio
    CS -> CS : messages.append(assistant_msg)
-    CS -> DB : save_message(session_id, "assistant", content)
-    CS -> DB : save_message(session_id, "tool_call", ...) ×N
+    CS -> DB : save_message(ws_id, "assistant", content)
+    CS -> DB : save_message(ws_id, "tool_call", ...) ×N

    == Tool Dispatch (if tool_calls) ==

@@ -136,7 +136,7 @@ group loop [while tool_calls present]

        loop for each result
            CS -> CS : messages.append({role: "tool", ...})
-            CS -> DB : save_message(session_id, "tool_result", ...)
+            CS -> DB : save_message(ws_id, "tool_result", ...)
        end

        opt user_feedback from approval
@@ -35,7 +35,7 @@ package "turnstone/sdk/ (Python)" {
        + stream_events(ws_id)
        + stream_global_events()
        + send_and_wait()
-        + list_sessions()
+        + list_saved_workstreams()
        + login() / logout()
        + health()
    }
@@ -13,23 +13,19 @@ skinparam class {

 ' -- Protocol --
 interface "StorageBackend" as SB <<protocol>> {
-    +register_session(session_id, title, node_id, ws_id)
-    +save_message(session_id, role, content, ...)
-    +load_session_messages(session_id) → list[dict]
-    +list_sessions(limit) → list
-    +delete_session(session_id) → bool
-    +prune_sessions(retention_days) → (int, int)
-    +resolve_session(alias_or_id) → str | None
-    +save_session_config(session_id, config)
-    +load_session_config(session_id) → dict
-    +set_session_alias(session_id, alias) → bool
-    +get_session_name(session_id) → str | None
-    +update_session_title(session_id, title)
+    +save_message(ws_id, role, content, ...)
+    +load_messages(ws_id) → list[dict]
    +register_workstream(ws_id, node_id, name, state)
    +update_workstream_state(ws_id, state)
    +update_workstream_name(ws_id, name)
+    +set_workstream_alias(ws_id, alias) → bool
+    +update_workstream_title(ws_id, title)
+    +resolve_workstream(alias_or_id) → str | None
    +delete_workstream(ws_id) → bool
+    +prune_workstreams(retention_days) → (int, int)
    +list_workstreams(node_id, limit) → list
+    +save_workstream_config(ws_id, config)
+    +load_workstream_config(ws_id) → dict
    +kv_get(key) → str | None
    +kv_set(key, value) → str | None
    +kv_delete(key) → bool
@@ -68,9 +64,8 @@ class "_schema.py" as Schema <<schema>> {
    +metadata: MetaData
    +memories: Table
    +conversations: Table
-    +sessions: Table (node_id, ws_id, user_id)
-    +workstreams: Table (node_id, user_id, state)
-    +session_config: Table
+    +workstreams: Table (node_id, alias, title, state)
+    +workstream_config: Table
    +users: Table (username, password_hash)
    +api_tokens: Table (token_hash, scopes)
    +channel_users: Table (channel_type)
@@ -106,14 +101,14 @@ class "_registry.py" as Registry {

 ' -- Facade --
 class "memory.py" as Facade <<facade>> {
-    +register_session()
    +save_message()
-    +load_session_messages()
+    +load_messages()
    +register_workstream()
    +update_workstream_state()
+    +save_workstream_config()
    +save_memory() / delete_memory()
    +search_memories()
-    +... (all 22 functions)
+    +... (all delegated functions)
    --
    Thin delegation to
    get_storage()
@@ -44,11 +44,15 @@ class "turnstone-channel" as ChannelService <<service>> {
    asyncio event loop
    Structured logging (structlog)
    --log-level, --log-format
+    --
+    POST /v1/api/notify (HTTP)
+    GET  /health
 }

 class "DiscordBot" as Bot <<service>> {
    +on_message(msg)
    +on_interaction(interaction)
+    +send(channel_id, content)
    +run(token)
    --
    discord.py Client
@@ -56,6 +60,7 @@ class "DiscordBot" as Bot <<service>> {
    Sends replies + embeds
    Creates threads for workstreams
    Renders approval buttons
+    escape_mentions() on send
 }

 class "ChannelRouter" as Router <<service>> {
@@ -109,6 +114,9 @@ class "turnstone-server" as Server <<server>> {
    --
    LLM execution + tool use
    SSE event stream
+    --
+    notify tool: _exec_notify()
+    ServiceTokenManager (JWT)
 }

 ' -- Storage --
@@ -134,6 +142,18 @@ class "channel_routes" as CR <<storage>> {
    to turnstone workstreams
 }

+class "services" as SVC <<storage>> {
+    service_type (PK)
+    service_id (PK)
+    url
+    last_heartbeat
+    created
+    --
+    Heartbeat every 30s
+    Stale after 120s
+    ON CONFLICT DO UPDATE
+}
+
 ' -- Relationships --
 Discord --> Bot : gateway\nevents
 Bot --> Router : on_message\non_interaction
@@ -157,6 +177,11 @@ Teams .[hidden]. Slack
 ChannelService --> Bot : creates + runs
 ChannelService --> Router : creates
 ChannelService --> Broker : creates
+ChannelService --> SVC : register / heartbeat /\nderegister
+
+' -- Notification path (direct HTTP, bypasses MQ) --
+Server --> ChannelService : POST /v1/api/notify\n(JWT: aud=turnstone-channel)
+Server --> SVC : list_services("channel",\nmax_age_seconds=120)

 ' -- Notes --
 note right of Bot
@@ -170,13 +195,13 @@ note right of Bot
  5. Broker.push_inbound(SendMessage)
  6. Bridge pops from Redis, drives server

-  **Session Resume (evicted workstreams)**
+  **Workstream Resume (evicted workstreams)**
  1. Stale route detected (no MQ owner)
-  2. Old session looked up via get_session_id_by_ws()
+  2. Existing ws_id reused directly from route
  3. CreateWorkstreamMessage sent with
-     resume_session=<old_session_id>
+     resume_ws=<ws_id>
  4. Server resumes atomically during creation
-  5. Bridge emits SessionResumedEvent → thread
+  5. Bridge emits WorkstreamResumedEvent → thread
 end note

 note right of Broker
@@ -208,4 +233,18 @@ note bottom of CU
  7. AuthResult scopes applied by server
 end note

+note bottom of SVC
+  **Notification Flow** (direct HTTP, bypasses MQ)
+  1. LLM calls notify tool → _prepare_notify()
+  2. _exec_notify() checks rate limit (5/turn)
+  3. Queries services table for healthy gateways
+  4. Mints JWT (aud: turnstone-channel) via
+     ServiceTokenManager
+  5. POSTs to first healthy gateway
+  6. Gateway validates JWT, resolves target
+  7. adapter.send() → Discord API
+  8. On failure: retry up to 3× (1s, 3s backoff)
+  9. SSRF: only http(s) URLs allowed
+end note
+
@enduml
@@ -0,0 +1,116 @@
+@startuml
+!theme plain
+title Turnstone — Notification Delivery Flow
+
+skinparam participant {
+    BackgroundColor<<server>> #FFE0B2
+    BackgroundColor<<storage>> #B3E5FC
+    BackgroundColor<<service>> #E8EAF6
+    BackgroundColor<<platform>> #E1BEE7
+}
+
+participant "ChatSession\n(turnstone-server)" as Session <<server>>
+participant "StorageBackend" as Storage <<storage>>
+participant "ServiceTokenManager" as STM <<server>>
+participant "Channel Gateway\n(_http.py)" as Gateway <<service>>
+participant "ChannelAdapter\n(Discord bot)" as Adapter <<service>>
+participant "Discord API" as Discord <<platform>>
+
+== Prepare Phase ==
+
+Session -> Session : _prepare_notify(call_id, args)
+note right
+  Validates:
+  - message (required, ≤2000 chars)
+  - target: username OR channel_type+channel_id
+  - no ambiguous targeting (both set)
+  - partial targeting errors
+end note
+
+== Execute Phase ==
+
+Session -> Session : _exec_notify(item)
+Session -> Session : check rate limit\n(≥5 per turn?)
+
+alt rate limit exceeded
+    Session --> Session : "Error: rate limit exceeded"
+end
+
+loop up to 3 attempts (retry delays: 1s, 3s)
+
+    Session -> Storage : list_services("channel",\nmax_age_seconds=120)
+    Storage --> Session : services[] (sorted by\nlast_heartbeat DESC)
+
+    alt no healthy services
+        Session -> Session : log.warning("notify.no_services")
+        Session -> Session : sleep(delay)
+    else services available
+
+        Session -> STM : bearer_header
+        note right
+          Lazy-init ServiceTokenManager
+          aud: turnstone-channel
+          scope: write
+          Auto-rotates 1h JWTs
+        end note
+        STM --> Session : Authorization: Bearer <jwt>
+
+        loop for each gateway (first-healthy)
+            Session -> Session : SSRF check:\nurl.startswith("http://"|"https://")
+
+            Session -> Gateway : POST /v1/api/notify\n+ Authorization header
+            Gateway -> Gateway : _check_auth()\nvalidate JWT (aud=turnstone-channel)\nor static token
+
+            alt auth failed
+                Gateway --> Session : 401 Unauthorized
+            else auth ok
+
+                alt username target
+                    Gateway -> Storage : get_user_by_username()
+                    Storage --> Gateway : user
+                    Gateway -> Storage : list_channel_users_by_user()
+                    Storage --> Gateway : linked channels
+                else direct target
+                    Gateway -> Gateway : use channel_type + channel_id
+                end
+
+                Gateway -> Adapter : send(channel_id, content)
+                note right
+                  escape_mentions() applied
+                  Chunked for 2000-char limit
+                end note
+                Adapter -> Discord : POST message
+                Discord --> Adapter : message_id
+                Adapter --> Gateway : message_id
+                Gateway --> Session : 200 {results: [{status: "sent"}]}
+
+                Session -> Session : _notify_count += 1
+                Session --> Session : "Notification sent successfully"
+                note right : Return — no further\ngateways tried
+            end
+        end
+
+        alt all gateways failed
+            Session -> Session : log.warning(\n"notify.all_gateways_failed")
+            Session -> Session : sleep(delay)
+        end
+
+    end
+end
+
+alt all retries exhausted
+    Session -> Session : log.warning("notify.delivery_failed")
+    Session --> Session : "Error: notification delivery failed"
+end
+
+== Service Registry (Background) ==
+
+note over Gateway, Storage
+  **Heartbeat Lifecycle**
+  1. Gateway startup: register_service("channel", id, url)
+  2. Every 30s: heartbeat_service("channel", id)
+  3. Shutdown: deregister_service("channel", id)
+  4. Stale after 120s (4 missed heartbeats)
+end note
+
+@enduml
@@ -0,0 +1,286 @@
+<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 1200 540" font-family="-apple-system, BlinkMacSystemFont, 'Segoe UI', Helvetica, Arial, sans-serif">
+  <defs>
+    <!-- Arrowhead markers -->
+    <marker id="arrow" viewBox="0 0 10 7" refX="10" refY="3.5" markerWidth="8" markerHeight="6" orient="auto-start-auto">
+      <path d="M 0 0 L 10 3.5 L 0 7 z" fill="#484f58"/>
+    </marker>
+    <marker id="arrow-blue" viewBox="0 0 10 7" refX="10" refY="3.5" markerWidth="8" markerHeight="6" orient="auto-start-auto">
+      <path d="M 0 0 L 10 3.5 L 0 7 z" fill="#58a6ff"/>
+    </marker>
+    <marker id="arrow-green" viewBox="0 0 10 7" refX="10" refY="3.5" markerWidth="8" markerHeight="6" orient="auto-start-auto">
+      <path d="M 0 0 L 10 3.5 L 0 7 z" fill="#3fb950"/>
+    </marker>
+    <marker id="arrow-orange" viewBox="0 0 10 7" refX="10" refY="3.5" markerWidth="8" markerHeight="6" orient="auto-start-auto">
+      <path d="M 0 0 L 10 3.5 L 0 7 z" fill="#f0883e"/>
+    </marker>
+    <marker id="arrow-coral" viewBox="0 0 10 7" refX="10" refY="3.5" markerWidth="8" markerHeight="6" orient="auto-start-auto">
+      <path d="M 0 0 L 10 3.5 L 0 7 z" fill="#f47067"/>
+    </marker>
+    <marker id="arrow-muted" viewBox="0 0 10 7" refX="10" refY="3.5" markerWidth="8" markerHeight="6" orient="auto-start-auto">
+      <path d="M 0 0 L 10 3.5 L 0 7 z" fill="#8b949e"/>
+    </marker>
+
+    <!-- Card shadow filter -->
+    <filter id="shadow" x="-4%" y="-4%" width="108%" height="112%">
+      <feDropShadow dx="0" dy="1" stdDeviation="2" flood-color="#000" flood-opacity="0.4"/>
+    </filter>
+  </defs>
+
+  <!-- Background -->
+  <rect width="1200" height="540" rx="8" fill="#0d1117"/>
+
+  <!-- Title -->
+  <text x="600" y="36" text-anchor="middle" fill="#e6edf3" font-size="15" font-weight="700" letter-spacing="3">TURNSTONE</text>
+  <text x="600" y="54" text-anchor="middle" fill="#8b949e" font-size="11" letter-spacing="1">SYSTEM ARCHITECTURE</text>
+
+  <!-- ==================== COLUMN HEADERS ==================== -->
+  <text x="90" y="86" text-anchor="middle" fill="#58a6ff" font-size="9" font-weight="600" letter-spacing="2">CLIENTS</text>
+  <text x="276" y="86" text-anchor="middle" fill="#3fb950" font-size="9" font-weight="600" letter-spacing="2">GATEWAYS</text>
+  <text x="480" y="86" text-anchor="middle" fill="#f0883e" font-size="9" font-weight="600" letter-spacing="2">MESSAGE QUEUE</text>
+  <text x="700" y="86" text-anchor="middle" fill="#f47067" font-size="9" font-weight="600" letter-spacing="2">CLUSTER NODES</text>
+  <text x="940" y="86" text-anchor="middle" fill="#f778ba" font-size="9" font-weight="600" letter-spacing="2">LLM PROVIDERS</text>
+
+  <!-- ==================== CLIENT BOXES ==================== -->
+  <!-- CLI -->
+  <g filter="url(#shadow)">
+    <rect x="30" y="108" width="120" height="46" rx="5" fill="#161b22" stroke="#30363d" stroke-width="1"/>
+    <rect x="30" y="108" width="120" height="5" rx="5" fill="#58a6ff"/>
+    <rect x="30" y="108" width="120" height="5" fill="#58a6ff"/>
+    <rect x="30" y="111" width="120" height="2" fill="#161b22"/>
+    <text x="90" y="130" text-anchor="middle" fill="#e6edf3" font-size="11" font-weight="600">CLI</text>
+    <text x="90" y="145" text-anchor="middle" fill="#8b949e" font-size="9">terminal REPL</text>
+  </g>
+
+  <!-- Browser UI -->
+  <g filter="url(#shadow)">
+    <rect x="30" y="174" width="120" height="46" rx="5" fill="#161b22" stroke="#30363d" stroke-width="1"/>
+    <rect x="30" y="174" width="120" height="5" rx="5" fill="#58a6ff"/>
+    <rect x="30" y="174" width="120" height="5" fill="#58a6ff"/>
+    <rect x="30" y="177" width="120" height="2" fill="#161b22"/>
+    <text x="90" y="196" text-anchor="middle" fill="#e6edf3" font-size="11" font-weight="600">Browser UI</text>
+    <text x="90" y="211" text-anchor="middle" fill="#8b949e" font-size="9">HTTP + SSE</text>
+  </g>
+
+  <!-- SDK / API -->
+  <g filter="url(#shadow)">
+    <rect x="30" y="244" width="120" height="46" rx="5" fill="#161b22" stroke="#30363d" stroke-width="1"/>
+    <rect x="30" y="244" width="120" height="5" rx="5" fill="#58a6ff"/>
+    <rect x="30" y="244" width="120" height="5" fill="#58a6ff"/>
+    <rect x="30" y="247" width="120" height="2" fill="#161b22"/>
+    <text x="90" y="266" text-anchor="middle" fill="#e6edf3" font-size="11" font-weight="600">SDK / API</text>
+    <text x="90" y="281" text-anchor="middle" fill="#8b949e" font-size="9">programmatic</text>
+  </g>
+
+  <!-- Discord / Slack -->
+  <g filter="url(#shadow)">
+    <rect x="30" y="314" width="120" height="46" rx="5" fill="#161b22" stroke="#30363d" stroke-width="1"/>
+    <rect x="30" y="314" width="120" height="5" rx="5" fill="#58a6ff"/>
+    <rect x="30" y="314" width="120" height="5" fill="#58a6ff"/>
+    <rect x="30" y="317" width="120" height="2" fill="#161b22"/>
+    <text x="90" y="336" text-anchor="middle" fill="#e6edf3" font-size="11" font-weight="600">Discord / Slack</text>
+    <text x="90" y="351" text-anchor="middle" fill="#8b949e" font-size="9">chat platforms</text>
+  </g>
+
+  <!-- ==================== GATEWAY BOXES ==================== -->
+  <!-- Console -->
+  <g filter="url(#shadow)">
+    <rect x="216" y="118" width="120" height="58" rx="5" fill="#161b22" stroke="#30363d" stroke-width="1"/>
+    <rect x="216" y="118" width="120" height="5" rx="5" fill="#3fb950"/>
+    <rect x="216" y="118" width="120" height="5" fill="#3fb950"/>
+    <rect x="216" y="121" width="120" height="2" fill="#161b22"/>
+    <text x="276" y="142" text-anchor="middle" fill="#e6edf3" font-size="11" font-weight="600">Console</text>
+    <text x="276" y="157" text-anchor="middle" fill="#8b949e" font-size="9">dashboard + proxy</text>
+    <text x="276" y="169" text-anchor="middle" fill="#8b949e" font-size="9">cluster management</text>
+  </g>
+
+  <!-- Channel Gateway -->
+  <g filter="url(#shadow)">
+    <rect x="216" y="292" width="120" height="58" rx="5" fill="#161b22" stroke="#30363d" stroke-width="1"/>
+    <rect x="216" y="292" width="120" height="5" rx="5" fill="#3fb950"/>
+    <rect x="216" y="292" width="120" height="5" fill="#3fb950"/>
+    <rect x="216" y="295" width="120" height="2" fill="#161b22"/>
+    <text x="276" y="316" text-anchor="middle" fill="#e6edf3" font-size="11" font-weight="600">Channel Gateway</text>
+    <text x="276" y="331" text-anchor="middle" fill="#8b949e" font-size="9">platform adapter</text>
+    <text x="276" y="343" text-anchor="middle" fill="#8b949e" font-size="9">Discord, Slack, ...</text>
+  </g>
+
+  <!-- ==================== REDIS MQ ==================== -->
+  <g filter="url(#shadow)">
+    <rect x="420" y="168" width="120" height="132" rx="5" fill="#161b22" stroke="#30363d" stroke-width="1"/>
+    <rect x="420" y="168" width="120" height="5" rx="5" fill="#f0883e"/>
+    <rect x="420" y="168" width="120" height="5" fill="#f0883e"/>
+    <rect x="420" y="171" width="120" height="2" fill="#161b22"/>
+    <text x="480" y="198" text-anchor="middle" fill="#e6edf3" font-size="11" font-weight="600">Redis MQ</text>
+    <line x1="438" y1="210" x2="522" y2="210" stroke="#30363d" stroke-width="1"/>
+    <text x="480" y="228" text-anchor="middle" fill="#8b949e" font-size="9">inbound queues</text>
+    <text x="480" y="243" text-anchor="middle" fill="#8b949e" font-size="9">event pub/sub</text>
+    <text x="480" y="258" text-anchor="middle" fill="#8b949e" font-size="9">node heartbeats</text>
+    <text x="480" y="273" text-anchor="middle" fill="#8b949e" font-size="9">workstream routing</text>
+    <text x="480" y="288" text-anchor="middle" fill="#8b949e" font-size="9">cluster state</text>
+  </g>
+
+  <!-- ==================== CLUSTER NODES ==================== -->
+  <!-- Cluster outline -->
+  <rect x="598" y="100" width="204" height="310" rx="8" fill="none" stroke="#30363d" stroke-width="1" stroke-dasharray="4,3"/>
+  <text x="700" y="422" text-anchor="middle" fill="#30363d" font-size="9" letter-spacing="1">CLUSTER</text>
+
+  <!-- Node A -->
+  <g filter="url(#shadow)">
+    <rect x="614" y="118" width="170" height="100" rx="5" fill="#161b22" stroke="#30363d" stroke-width="1"/>
+    <rect x="614" y="118" width="170" height="5" rx="5" fill="#f47067"/>
+    <rect x="614" y="118" width="170" height="5" fill="#f47067"/>
+    <rect x="614" y="121" width="170" height="2" fill="#161b22"/>
+    <text x="699" y="142" text-anchor="middle" fill="#e6edf3" font-size="11" font-weight="600">Node A</text>
+    <line x1="632" y1="152" x2="766" y2="152" stroke="#30363d" stroke-width="1"/>
+    <!-- Bridge -->
+    <rect x="626" y="162" width="70" height="28" rx="3" fill="#1c2128" stroke="#30363d" stroke-width="1"/>
+    <text x="661" y="180" text-anchor="middle" fill="#8b949e" font-size="9">bridge</text>
+    <!-- Server -->
+    <rect x="704" y="162" width="70" height="28" rx="3" fill="#1c2128" stroke="#30363d" stroke-width="1"/>
+    <text x="739" y="180" text-anchor="middle" fill="#8b949e" font-size="9">server</text>
+    <!-- Arrow bridge to server -->
+    <line x1="696" y1="176" x2="702" y2="176" stroke="#484f58" stroke-width="1" marker-end="url(#arrow)"/>
+    <!-- Tools label -->
+    <text x="699" y="206" text-anchor="middle" fill="#484f58" font-size="8">14 tools + MCP</text>
+  </g>
+
+  <!-- Node B -->
+  <g filter="url(#shadow)">
+    <rect x="614" y="238" width="170" height="100" rx="5" fill="#161b22" stroke="#30363d" stroke-width="1"/>
+    <rect x="614" y="238" width="170" height="5" rx="5" fill="#f47067"/>
+    <rect x="614" y="238" width="170" height="5" fill="#f47067"/>
+    <rect x="614" y="241" width="170" height="2" fill="#161b22"/>
+    <text x="699" y="262" text-anchor="middle" fill="#e6edf3" font-size="11" font-weight="600">Node B</text>
+    <line x1="632" y1="272" x2="766" y2="272" stroke="#30363d" stroke-width="1"/>
+    <!-- Bridge -->
+    <rect x="626" y="282" width="70" height="28" rx="3" fill="#1c2128" stroke="#30363d" stroke-width="1"/>
+    <text x="661" y="300" text-anchor="middle" fill="#8b949e" font-size="9">bridge</text>
+    <!-- Server -->
+    <rect x="704" y="282" width="70" height="28" rx="3" fill="#1c2128" stroke="#30363d" stroke-width="1"/>
+    <text x="739" y="300" text-anchor="middle" fill="#8b949e" font-size="9">server</text>
+    <!-- Arrow bridge to server -->
+    <line x1="696" y1="296" x2="702" y2="296" stroke="#484f58" stroke-width="1" marker-end="url(#arrow)"/>
+    <!-- Tools label -->
+    <text x="699" y="326" text-anchor="middle" fill="#484f58" font-size="8">14 tools + MCP</text>
+  </g>
+
+  <!-- ==================== LLM PROVIDERS ==================== -->
+  <!-- OpenAI -->
+  <g filter="url(#shadow)">
+    <rect x="870" y="130" width="140" height="46" rx="5" fill="#161b22" stroke="#30363d" stroke-width="1"/>
+    <rect x="870" y="130" width="140" height="5" rx="5" fill="#f778ba"/>
+    <rect x="870" y="130" width="140" height="5" fill="#f778ba"/>
+    <rect x="870" y="133" width="140" height="2" fill="#161b22"/>
+    <text x="940" y="153" text-anchor="middle" fill="#e6edf3" font-size="11" font-weight="600">OpenAI</text>
+    <text x="940" y="167" text-anchor="middle" fill="#8b949e" font-size="9">GPT-5, o-series</text>
+  </g>
+
+  <!-- Anthropic -->
+  <g filter="url(#shadow)">
+    <rect x="870" y="196" width="140" height="46" rx="5" fill="#161b22" stroke="#30363d" stroke-width="1"/>
+    <rect x="870" y="196" width="140" height="5" rx="5" fill="#f778ba"/>
+    <rect x="870" y="196" width="140" height="5" fill="#f778ba"/>
+    <rect x="870" y="199" width="140" height="2" fill="#161b22"/>
+    <text x="940" y="219" text-anchor="middle" fill="#e6edf3" font-size="11" font-weight="600">Anthropic</text>
+    <text x="940" y="233" text-anchor="middle" fill="#8b949e" font-size="9">Claude 4.5 / 4.6</text>
+  </g>
+
+  <!-- Local / vLLM -->
+  <g filter="url(#shadow)">
+    <rect x="870" y="262" width="140" height="46" rx="5" fill="#161b22" stroke="#30363d" stroke-width="1"/>
+    <rect x="870" y="262" width="140" height="5" rx="5" fill="#f778ba"/>
+    <rect x="870" y="262" width="140" height="5" fill="#f778ba"/>
+    <rect x="870" y="265" width="140" height="2" fill="#161b22"/>
+    <text x="940" y="285" text-anchor="middle" fill="#e6edf3" font-size="11" font-weight="600">Local / vLLM</text>
+    <text x="940" y="299" text-anchor="middle" fill="#8b949e" font-size="9">llama.cpp, NIM</text>
+  </g>
+
+  <!-- ==================== STORAGE ==================== -->
+  <g filter="url(#shadow)">
+    <rect x="614" y="450" width="170" height="52" rx="5" fill="#161b22" stroke="#30363d" stroke-width="1"/>
+    <rect x="614" y="450" width="170" height="5" rx="5" fill="#bc8cff"/>
+    <rect x="614" y="450" width="170" height="5" fill="#bc8cff"/>
+    <rect x="614" y="453" width="170" height="2" fill="#161b22"/>
+    <text x="699" y="476" text-anchor="middle" fill="#e6edf3" font-size="11" font-weight="600">PostgreSQL / SQLite</text>
+    <text x="699" y="492" text-anchor="middle" fill="#8b949e" font-size="9">conversations, memory, auth</text>
+  </g>
+  <text x="699" y="444" text-anchor="middle" fill="#bc8cff" font-size="9" font-weight="600" letter-spacing="2">STORAGE</text>
+
+  <!-- ==================== CONNECTION LINES ==================== -->
+
+  <!-- CLIENT -> GATEWAY connections -->
+  <!-- Browser -> Console -->
+  <line x1="150" y1="197" x2="214" y2="155" stroke="#58a6ff" stroke-width="1.2" stroke-opacity="0.6" marker-end="url(#arrow-blue)"/>
+  <!-- Discord -> Channel -->
+  <line x1="150" y1="337" x2="214" y2="325" stroke="#58a6ff" stroke-width="1.2" stroke-opacity="0.6" marker-end="url(#arrow-blue)"/>
+
+  <!-- CLI -> direct to Node A server (top path, curved) -->
+  <path d="M 150 131 C 200 131, 200 100, 400 100 L 400 100 C 500 100, 570 140, 612 168" stroke="#58a6ff" stroke-width="1.2" stroke-opacity="0.5" fill="none" stroke-dasharray="6,3" marker-end="url(#arrow-blue)"/>
+  <text x="370" y="96" fill="#484f58" font-size="8" text-anchor="middle">direct</text>
+
+  <!-- SDK -> Redis (direct push) -->
+  <line x1="150" y1="267" x2="418" y2="240" stroke="#58a6ff" stroke-width="1.2" stroke-opacity="0.6" marker-end="url(#arrow-blue)"/>
+
+  <!-- GATEWAY -> REDIS connections -->
+  <!-- Console -> Redis -->
+  <line x1="336" y1="160" x2="418" y2="200" stroke="#3fb950" stroke-width="1.2" stroke-opacity="0.6" marker-end="url(#arrow-green)"/>
+  <!-- Channel -> Redis -->
+  <line x1="336" y1="318" x2="418" y2="272" stroke="#3fb950" stroke-width="1.2" stroke-opacity="0.6" marker-end="url(#arrow-green)"/>
+
+  <!-- REDIS -> NODE connections -->
+  <!-- Redis -> Node A bridge -->
+  <line x1="540" y1="210" x2="624" y2="176" stroke="#f0883e" stroke-width="1.2" stroke-opacity="0.6" marker-end="url(#arrow-orange)"/>
+  <!-- Redis -> Node B bridge -->
+  <line x1="540" y1="260" x2="624" y2="296" stroke="#f0883e" stroke-width="1.2" stroke-opacity="0.6" marker-end="url(#arrow-orange)"/>
+
+  <!-- Console -> Node (proxy, dashed) -->
+  <path d="M 336 147 C 380 130, 500 108, 612 145" stroke="#3fb950" stroke-width="1" stroke-opacity="0.4" fill="none" stroke-dasharray="4,3" marker-end="url(#arrow-green)"/>
+  <text x="468" y="120" fill="#484f58" font-size="8" text-anchor="middle">proxy</text>
+
+  <!-- NODE -> LLM connections -->
+  <!-- Node A -> LLM providers -->
+  <line x1="784" y1="168" x2="868" y2="155" stroke="#f47067" stroke-width="1.2" stroke-opacity="0.5" marker-end="url(#arrow-coral)"/>
+  <line x1="784" y1="176" x2="868" y2="219" stroke="#f47067" stroke-width="1.2" stroke-opacity="0.3"/>
+  <line x1="784" y1="180" x2="868" y2="282" stroke="#f47067" stroke-width="1.2" stroke-opacity="0.2"/>
+
+  <!-- Node B -> LLM providers -->
+  <line x1="784" y1="288" x2="868" y2="163" stroke="#f47067" stroke-width="1.2" stroke-opacity="0.2"/>
+  <line x1="784" y1="296" x2="868" y2="222" stroke="#f47067" stroke-width="1.2" stroke-opacity="0.3"/>
+  <line x1="784" y1="300" x2="868" y2="288" stroke="#f47067" stroke-width="1.2" stroke-opacity="0.5" marker-end="url(#arrow-coral)"/>
+
+  <!-- NODE -> STORAGE connections -->
+  <line x1="680" y1="338" x2="680" y2="448" stroke="#bc8cff" stroke-width="1.2" stroke-opacity="0.4" stroke-dasharray="4,3" marker-end="url(#arrow-muted)"/>
+  <line x1="718" y1="218" x2="718" y2="236" stroke="#484f58" stroke-width="1" stroke-opacity="0.3" stroke-dasharray="2,2"/>
+
+  <!-- Extensibility hint -->
+  <text x="699" y="392" text-anchor="middle" fill="#30363d" font-size="10">...</text>
+
+  <!-- Event flow: Bridges -> Redis (dashed, bidirectional feel) -->
+  <line x1="624" y1="186" x2="542" y2="220" stroke="#f0883e" stroke-width="1" stroke-opacity="0.3" stroke-dasharray="3,3"/>
+  <line x1="624" y1="286" x2="542" y2="250" stroke="#f0883e" stroke-width="1" stroke-opacity="0.3" stroke-dasharray="3,3"/>
+  <text x="574" y="242" fill="#484f58" font-size="7" text-anchor="middle">events</text>
+
+  <!-- ==================== FLOW LABELS ==================== -->
+  <!-- Interactive flow label -->
+  <rect x="30" y="395" width="10" height="10" rx="2" fill="none" stroke="#58a6ff" stroke-width="1.5" stroke-dasharray="3,2"/>
+  <text x="46" y="404" fill="#8b949e" font-size="9">interactive (direct)</text>
+
+  <!-- Queue flow label -->
+  <rect x="160" y="395" width="10" height="10" rx="2" fill="none" stroke="#f0883e" stroke-width="1.5"/>
+  <text x="176" y="404" fill="#8b949e" font-size="9">queue-driven</text>
+
+  <!-- Proxy/event label -->
+  <rect x="275" y="395" width="10" height="10" rx="2" fill="none" stroke="#3fb950" stroke-width="1.5" stroke-dasharray="3,2"/>
+  <text x="291" y="404" fill="#8b949e" font-size="9">proxy / events</text>
+
+  <!-- ==================== BOTTOM DETAILS ==================== -->
+  <line x1="30" y1="430" x2="1170" y2="430" stroke="#21262d" stroke-width="1"/>
+
+  <!-- Routing rules at bottom, left-aligned -->
+  <text x="44" y="456" fill="#30363d" font-size="9" font-weight="600" letter-spacing="1">ROUTING</text>
+  <circle cx="44" cy="474" r="3" fill="#f47067" opacity="0.6"/>
+  <text x="54" y="477" fill="#484f58" font-size="9">target_node set  &#x2192;  route to specific node queue</text>
+  <circle cx="44" cy="494" r="3" fill="#f0883e" opacity="0.6"/>
+  <text x="54" y="497" fill="#484f58" font-size="9">ws_id set  &#x2192;  route to owning node</text>
+  <circle cx="44" cy="514" r="3" fill="#58a6ff" opacity="0.6"/>
+  <text x="54" y="517" fill="#484f58" font-size="9">neither  &#x2192;  shared queue, any node picks up</text></svg>
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:47b106bbcb1041fe4122007065c6cc85605348b42fc7140287194cf25e42e095
-size 318036
+oid sha256:6049cc0b07480df88d0d93aa977a1e97f64b41588325ff41d98be0e39431fc5c
+size 431712
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f55e177e0838a16d9bc4f07b162b4b6a966cc596c9d0a022d35a3c84f23e7b02
+size 221452
@@ -28,6 +28,8 @@ Console dashboard: http://localhost:8090
 | `bridge` | — | default | Redis-to-HTTP bridge (multi-node routing) |
 | `console` | 8090 | default | Cluster dashboard |
 | `channel` | — | production | Channel gateway (Discord, Slack, etc.) |
+| `server-1`…`server-10` | — | cluster | 10-node server fleet (PostgreSQL required) |
+| `bridge-1`…`bridge-10` | — | cluster | Matching bridge fleet |
 | `sim` | — | sim | Multi-node cluster simulator |

 ## Profiles
@@ -44,6 +46,12 @@ docker compose up
 docker compose --profile production up
 ```

+**Cluster** — 10-node server/bridge fleet sharing PostgreSQL and Redis. Access all nodes via the console at `:8090`. Requires `POSTGRES_PASSWORD`:
+
+```bash
+docker compose --profile cluster up
+```
+
 **Sim** — adds the simulator. Can run alongside the full stack or standalone with just Redis and the console:

 ```bash
@@ -135,13 +143,13 @@ The channel service runs in the `production` profile. When `TURNSTONE_DISCORD_TO

 ## Scaling

-Scale to multiple server/bridge pairs:
+For multi-node testing, use the `cluster` profile which provides 10 dedicated server+bridge pairs with unique node IDs (`node-1` through `node-10`), resource limits, and shared PostgreSQL:

 ```bash
-docker compose up --scale server=3 --scale bridge=3
+POSTGRES_PASSWORD=secret docker compose --profile cluster up
 ```

-Each bridge auto-generates a unique node ID from its container hostname. When scaling `server`, remove the host port mapping (or use a reverse proxy) to avoid port conflicts.
+The default `server` and `bridge` also run alongside the cluster nodes (11 total). All nodes are accessible via the console dashboard at `:8090`.

 ## Volumes

@@ -78,7 +78,7 @@ Both `TurnstoneServer` (sync) and `AsyncTurnstoneServer` (async) expose:
 | **Streaming** | `stream_events(ws_id)` | `Iterator[ServerEvent]` |
 | | `stream_global_events()` | `Iterator[ServerEvent]` |
 | **High-level** | `send_and_wait(message, ws_id, *, timeout, on_event)` | `TurnResult` |
-| **Sessions** | `list_sessions()` | `ListSessionsResponse` |
+| **Saved** | `list_saved_workstreams()` | `ListSavedWorkstreamsResponse` |
 | **Auth** | `login(username=..., password=...)` | `AuthLoginResponse` |
 | | `login(token="ts_xxx")` | `AuthLoginResponse` |
 | | `logout()` | `StatusResponse` |
@@ -96,6 +96,12 @@ Both `TurnstoneConsole` (sync) and `AsyncTurnstoneConsole` (async) expose:
 | | `workstreams(*, state, node, search, sort, page, per_page)` | `ClusterWorkstreamsResponse` |
 | | `node_detail(node_id)` | `NodeDetailResponse` |
 | | `create_workstream(*, node_id, name, model, initial_message)` | `ConsoleCreateWsResponse` |
+| **Schedules** | `list_schedules()` | `ListSchedulesResponse` |
+| | `create_schedule(*, name, schedule_type, initial_message, ...)` | `ScheduleInfo` |
+| | `get_schedule(task_id)` | `ScheduleInfo` |
+| | `update_schedule(task_id, *, name=..., enabled=..., ...)` | `ScheduleInfo` |
+| | `delete_schedule(task_id)` | `StatusResponse` |
+| | `list_schedule_runs(task_id, *, limit=50)` | `ListScheduleRunsResponse` |
 | **Streaming** | `stream_cluster_events()` | `Iterator[ClusterEvent]` |
 | **Auth** | `login(username=..., password=...)` / `login(token="ts_xxx")` | `AuthLoginResponse` |
 | | `logout()` | `StatusResponse` |
@@ -73,7 +73,7 @@ Scopes are hierarchical — higher scopes imply all lower ones.

 | Scope | Grants | Implies |
 |-------|--------|---------|
-| `read` | View workstreams, sessions, history | — |
+| `read` | View workstreams, saved workstreams, history | — |
 | `write` | Send messages, create/close workstreams | `read` |
 | `approve` | Approve tool calls, admin endpoints | `read`, `write` |

@@ -354,15 +354,22 @@ provided, that static token is used instead.
 The bridge and console collector use `ServiceTokenManager` for
 auto-rotating JWTs when communicating with server nodes:

-| Service | Identity | Scope | Purpose |
-|---------|----------|-------|---------|
-| Bridge | `bridge` | `approve` | Tool approval proxy, message relay |
-| Console collector | `console-collector` | `read` | Node health polling |
-| Console proxy | `console-proxy` | `write` | Proxied API calls |
+| Service | Identity | Scope | Audience | Purpose |
+|---------|----------|-------|----------|---------|
+| Bridge | `bridge` | `approve` | `turnstone-server` | Tool approval proxy, message relay |
+| Console collector | `console-collector` | `read` | `turnstone-server` | Node health polling |
+| Console proxy | `console-proxy` | `write` | `turnstone-server` | Proxied API calls |
+| Channel notify | `system` | `write` | `turnstone-channel` | Notification delivery to channel gateway |

-All service tokens use `aud: turnstone-server` and 1-hour expiry with
-automatic refresh.  The bridge injects auth headers per-request via httpx
-event hooks to ensure rotated tokens are picked up on SSE reconnects.
+Service tokens use 1-hour expiry with automatic refresh via
+`ServiceTokenManager`.  The bridge injects auth headers per-request via
+httpx event hooks to ensure rotated tokens are picked up on SSE
+reconnects.
+
+Note that the channel gateway uses a distinct JWT audience
+(`turnstone-channel`) from the server (`turnstone-server`) and console
+(`turnstone-console`).  A server-scoped JWT cannot authenticate to the
+channel gateway endpoint, and vice versa.

 ---

@@ -1,6 +1,6 @@
 # Tools Reference

-turnstone exposes 14 built-in tools plus any number of external MCP tools to the
+turnstone exposes 15 built-in tools plus any number of external MCP tools to the
 LLM via the OpenAI function-calling interface. Built-in tools are defined as JSON
 files under `turnstone/tools/` and loaded at startup by `turnstone/core/tools.py`.
 MCP tools are discovered from configured MCP servers at startup by
@@ -46,7 +46,7 @@ schema plus turnstone-specific metadata keys:

 | Name                | Description |
 |---------------------|-------------|
-| `TOOLS`             | All 14 tool definitions (sent to the model). |
+| `TOOLS`             | All 15 tool definitions (sent to the model). |
 | `AGENT_TOOLS`       | Tools with `agent: true` -- available to plan sub-agents. Read-only tools. |
 | `TASK_AGENT_TOOLS`  | Tools with `task_agent: true` -- available to task sub-agents. Includes write operations. |
 | `AGENT_AUTO_TOOLS`  | Set of tool names with `auto_approve: true` -- no user confirmation needed. |
@@ -113,6 +113,7 @@ Each item's `execute` callable is invoked:
 - `remember` -- writes to persistent memory database (lightweight, always auto-approved)
 - `recall` -- reads from persistent memory database
 - `forget` -- deletes from persistent memory database (lightweight, always auto-approved)
+- `notify` -- sends notifications to linked channels (time-sensitive, auto-approved for urgency)

 **Requires user confirmation** (write operations, network access, side effects):
 - `bash` -- arbitrary command execution
@@ -162,6 +163,7 @@ Every tool defines a `primary_key`. The mapping is:
 | `remember`   | `key`       |
 | `recall`     | `query`     |
 | `forget`     | `key`       |
+| `notify`     | `message`   |

 ---

@@ -335,7 +337,7 @@ Plan before implementing -- an autonomous agent explores the codebase and writes
 |-----------|--------|----------|-------------|
 | `prompt`  | string | yes      | What to plan -- the goal, constraints, and scope. |

- **What it does**: Spawns a planning sub-agent with `AGENT_TOOLS` (read-only tools: `read_file`, `search`, `math`, `man`, `web_fetch`, `web_search`). The agent explores the codebase and writes a structured plan to `.plan-<session_id>.md` (unique per session, so concurrent workstreams never collide). If the `plan` tool has been called before in the same session, the prior plan is passed to the agent as context so it refines rather than restarts. After completion, the user is prompted to review and can accept, reject, or annotate the plan.
+- **What it does**: Spawns a planning sub-agent with `AGENT_TOOLS` (read-only tools: `read_file`, `search`, `math`, `man`, `web_fetch`, `web_search`). The agent explores the codebase and writes a structured plan to `.plan-<ws_id>.md` (unique per workstream, so concurrent workstreams never collide). If the `plan` tool has been called before in the same session, the prior plan is passed to the agent as context so it refines rather than restarts. After completion, the user is prompted to review and can accept, reject, or annotate the plan.
 - **Auto-approve**: No -- requires user confirmation, plus post-execution review gate.
 - **Agent availability**: Not available to sub-agents (top-level only).

@@ -387,6 +389,33 @@ Remove a persistent memory by key.

 ---

+## Notifications
+
+### notify
+
+Send a notification to a user or channel on an external platform.
+
+| Parameter      | Type   | Required | Description |
+|----------------|--------|----------|-------------|
+| `message`      | string | yes      | Notification content (plain text, max 2000 chars). |
+| `username`     | string | no       | Turnstone username — sends to all linked channels. |
+| `channel_type` | string | no       | Platform for direct targeting (`discord`). |
+| `channel_id`   | string | no       | Platform-specific channel or user ID for direct targeting. |
+| `title`        | string | no       | Optional short title (rendered as bold prefix). |
+
+Provide either `username` for user-based targeting or `channel_type` +
+`channel_id` for direct targeting. Do not combine both.
+
+- **What it does**: Sends a notification via the channel gateway's HTTP endpoint (`POST /v1/api/notify`). The server queries the `services` table for healthy channel gateways, authenticates with a service JWT (`aud: turnstone-channel`), and delivers to the first healthy gateway. On failure, retries up to 2 additional times with backoff (1s, 3s). Rate-limited to 5 notifications per turn (counter only increments on success).
+- **Auto-approve**: Yes — notifications are time-sensitive and auto-approved so the model can alert users urgently.
+- **Agent availability**: `agent` and `task_agent`.
+
+> See [Channel Integrations: Notifications](channels.md#notifications)
+> for the full delivery flow, service registry details, and security
+> measures.
+
+---
+
 ## Summary Table

 | Tool         | Category   | Auto-approve | agent | task_agent | primary_key |
@@ -405,6 +434,7 @@ Remove a persistent memory by key.
 | `remember`   | Memory     | Yes          | No    | No         | `key`       |
 | `recall`     | Memory     | Yes          | No    | No         | `query`     |
 | `forget`     | Memory     | Yes          | No    | No         | `key`       |
+| `notify`     | Notify     | Yes          | Yes   | Yes        | `message`   |

 ---

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"

 [project]
 name = "turnstone"
-version = "0.4.0"
+version = "0.4.3"
 description = "Multi-node AI orchestration platform with tool use, agent routing, and cluster simulation."
 readme = "README.md"
 license = "BUSL-1.1"
@@ -43,10 +43,10 @@ Repository = "https://github.com/turnstonelabs/turnstone"
 Issues = "https://github.com/turnstonelabs/turnstone/issues"

 [project.optional-dependencies]
-test = ["pytest>=9.0", "pytest-cov>=6.0"]
+test = ["pytest>=9.0", "pytest-cov>=6.0", "croniter>=3.0"]
 dev = ["ruff>=0.9", "mypy>=1.14", "types-redis>=4.6"]
 mq = ["redis>=7.2"]
-console = ["redis>=7.2"]
+console = ["redis>=7.2", "croniter>=3.0"]
 sim = ["redis>=7.2"]
 anthropic = ["anthropic>=0.39"]
 postgres = ["psycopg[binary]>=3.2"]
@@ -150,6 +150,10 @@ ignore_missing_imports = true
 module = ["discord", "discord.*"]
 ignore_missing_imports = true

+[[tool.mypy.overrides]]
+module = ["croniter", "croniter.*"]
+ignore_missing_imports = true
+
 [[tool.mypy.overrides]]
 module = ["turnstone.channels.discord.*"]
 disallow_subclassing_any = false
@@ -2,7 +2,7 @@
  "openapi": "3.1.0",
  "info": {
    "title": "turnstone Server API",
-    "version": "0.3.0",
+    "version": "0.4.2",
    "description": "Single-node workstream management, chat interaction, and real-time streaming."
  },
  "paths": {
@@ -365,12 +365,12 @@
        }
      }
    },
-    "/v1/api/sessions": {
+    "/v1/api/workstreams/saved": {
      "get": {
-        "summary": "List saved sessions",
-        "operationId": "v1_api_sessions_get",
+        "summary": "List saved workstreams",
+        "operationId": "v1_api_workstreams_saved_get",
        "tags": [
-          "Sessions"
+          "Workstreams"
        ],
        "responses": {
          "200": {
@@ -378,7 +378,7 @@
            "content": {
              "application/json": {
                "schema": {
-                  "$ref": "#/components/schemas/ListSessionsResponse"
+                  "$ref": "#/components/schemas/ListSavedWorkstreamsResponse"
                }
              }
            }
@@ -427,6 +427,88 @@
        }
      }
    },
+    "/v1/api/auth/setup": {
+      "post": {
+        "summary": "Create first admin user",
+        "operationId": "v1_api_auth_setup_post",
+        "tags": [
+          "Auth"
+        ],
+        "requestBody": {
+          "required": true,
+          "content": {
+            "application/json": {
+              "schema": {
+                "$ref": "#/components/schemas/AuthSetupRequest"
+              }
+            }
+          }
+        },
+        "responses": {
+          "200": {
+            "description": "Success",
+            "content": {
+              "application/json": {
+                "schema": {
+                  "$ref": "#/components/schemas/AuthSetupResponse"
+                }
+              }
+            }
+          },
+          "400": {
+            "description": "Error 400",
+            "content": {
+              "application/json": {
+                "schema": {
+                  "$ref": "#/components/schemas/ErrorResponse"
+                }
+              }
+            }
+          },
+          "409": {
+            "description": "Error 409",
+            "content": {
+              "application/json": {
+                "schema": {
+                  "$ref": "#/components/schemas/ErrorResponse"
+                }
+              }
+            }
+          },
+          "503": {
+            "description": "Error 503",
+            "content": {
+              "application/json": {
+                "schema": {
+                  "$ref": "#/components/schemas/ErrorResponse"
+                }
+              }
+            }
+          }
+        }
+      }
+    },
+    "/v1/api/auth/status": {
+      "get": {
+        "summary": "Return auth state",
+        "operationId": "v1_api_auth_status_get",
+        "tags": [
+          "Auth"
+        ],
+        "responses": {
+          "200": {
+            "description": "Success",
+            "content": {
+              "application/json": {
+                "schema": {
+                  "$ref": "#/components/schemas/AuthStatusResponse"
+                }
+              }
+            }
+          }
+        }
+      }
+    },
    "/v1/api/auth/logout": {
      "post": {
        "summary": "Clear auth cookie",
@@ -503,17 +585,27 @@
        "type": "object"
      },
      "AuthLoginRequest": {
-        "description": "POST /v1/api/auth/login request body.",
+        "description": "POST /v1/api/auth/login request body.\n\nEither username+password or token must be provided.",
        "properties": {
+          "username": {
+            "default": "",
+            "description": "Login username",
+            "title": "Username",
+            "type": "string"
+          },
+          "password": {
+            "default": "",
+            "description": "Login password",
+            "title": "Password",
+            "type": "string"
+          },
          "token": {
-            "description": "Bearer token to authenticate",
+            "default": "",
+            "description": "Legacy: bearer token to authenticate",
            "title": "Token",
            "type": "string"
          }
        },
-        "required": [
-          "token"
-        ],
        "title": "AuthLoginRequest",
        "type": "object"
      },
@@ -525,14 +617,35 @@
            "title": "Status",
            "type": "string"
          },
+          "user_id": {
+            "default": "",
+            "description": "Authenticated user ID",
+            "title": "User Id",
+            "type": "string"
+          },
          "role": {
-            "description": "Assigned role",
+            "description": "Legacy role",
            "examples": [
              "full",
              "read"
            ],
            "title": "Role",
            "type": "string"
+          },
+          "scopes": {
+            "default": "",
+            "description": "Comma-separated scopes",
+            "examples": [
+              "read,write,approve"
+            ],
+            "title": "Scopes",
+            "type": "string"
+          },
+          "jwt": {
+            "default": "",
+            "description": "JWT session token (if JWT auth is configured)",
+            "title": "Jwt",
+            "type": "string"
          }
        },
        "required": [
@@ -541,6 +654,97 @@
        "title": "AuthLoginResponse",
        "type": "object"
      },
+      "AuthSetupRequest": {
+        "description": "POST /v1/api/auth/setup request body.",
+        "properties": {
+          "username": {
+            "description": "Login username (1-64 ASCII characters)",
+            "title": "Username",
+            "type": "string"
+          },
+          "display_name": {
+            "description": "Display name",
+            "title": "Display Name",
+            "type": "string"
+          },
+          "password": {
+            "description": "Password (minimum 8 characters)",
+            "title": "Password",
+            "type": "string"
+          }
+        },
+        "required": [
+          "username",
+          "display_name",
+          "password"
+        ],
+        "title": "AuthSetupRequest",
+        "type": "object"
+      },
+      "AuthSetupResponse": {
+        "description": "POST /v1/api/auth/setup success response.",
+        "properties": {
+          "status": {
+            "default": "ok",
+            "title": "Status",
+            "type": "string"
+          },
+          "user_id": {
+            "title": "User Id",
+            "type": "string"
+          },
+          "username": {
+            "title": "Username",
+            "type": "string"
+          },
+          "role": {
+            "default": "full",
+            "title": "Role",
+            "type": "string"
+          },
+          "scopes": {
+            "default": "approve,read,write",
+            "title": "Scopes",
+            "type": "string"
+          },
+          "jwt": {
+            "default": "",
+            "description": "JWT session token",
+            "title": "Jwt",
+            "type": "string"
+          }
+        },
+        "required": [
+          "user_id",
+          "username"
+        ],
+        "title": "AuthSetupResponse",
+        "type": "object"
+      },
+      "AuthStatusResponse": {
+        "description": "GET /v1/api/auth/status response.",
+        "properties": {
+          "auth_enabled": {
+            "title": "Auth Enabled",
+            "type": "boolean"
+          },
+          "has_users": {
+            "title": "Has Users",
+            "type": "boolean"
+          },
+          "setup_required": {
+            "title": "Setup Required",
+            "type": "boolean"
+          }
+        },
+        "required": [
+          "auth_enabled",
+          "has_users",
+          "setup_required"
+        ],
+        "title": "AuthStatusResponse",
+        "type": "object"
+      },
      "SendRequest": {
        "properties": {
          "message": {
@@ -677,6 +881,12 @@
            "description": "Auto-approve all tool calls",
            "title": "Auto Approve",
            "type": "boolean"
+          },
+          "resume_ws": {
+            "default": "",
+            "description": "Workstream ID to resume atomically during creation (empty = fresh start)",
+            "title": "Resume Ws",
+            "type": "string"
          }
        },
        "title": "CreateWorkstreamRequest",
@@ -693,6 +903,18 @@
            "description": "Assigned workstream name",
            "title": "Name",
            "type": "string"
+          },
+          "resumed": {
+            "default": false,
+            "description": "Whether a previous workstream was resumed",
+            "title": "Resumed",
+            "type": "boolean"
+          },
+          "message_count": {
+            "default": 0,
+            "description": "Number of messages in the resumed workstream",
+            "title": "Message Count",
+            "type": "integer"
          }
        },
        "required": [
@@ -745,18 +967,6 @@
          "state": {
            "title": "State",
            "type": "string"
-          },
-          "session_id": {
-            "anyOf": [
-              {
-                "type": "string"
-              },
-              {
-                "type": "null"
-              }
-            ],
-            "default": null,
-            "title": "Session Id"
          }
        },
        "required": [
@@ -837,18 +1047,6 @@
            "title": "State",
            "type": "string"
          },
-          "session_id": {
-            "anyOf": [
-              {
-                "type": "string"
-              },
-              {
-                "type": "null"
-              }
-            ],
-            "default": null,
-            "title": "Session Id"
-          },
          "title": {
            "default": "",
            "title": "Title",
@@ -903,26 +1101,26 @@
        "title": "DashboardWorkstream",
        "type": "object"
      },
-      "ListSessionsResponse": {
+      "ListSavedWorkstreamsResponse": {
        "properties": {
-          "sessions": {
+          "workstreams": {
            "items": {
-              "$ref": "#/components/schemas/SessionInfo"
+              "$ref": "#/components/schemas/SavedWorkstreamInfo"
            },
-            "title": "Sessions",
+            "title": "Workstreams",
            "type": "array"
          }
        },
        "required": [
-          "sessions"
+          "workstreams"
        ],
-        "title": "ListSessionsResponse",
+        "title": "ListSavedWorkstreamsResponse",
        "type": "object"
      },
-      "SessionInfo": {
+      "SavedWorkstreamInfo": {
        "properties": {
-          "session_id": {
-            "title": "Session Id",
+          "ws_id": {
+            "title": "Ws Id",
            "type": "string"
          },
          "alias": {
@@ -963,12 +1161,12 @@
          }
        },
        "required": [
-          "session_id",
+          "ws_id",
          "created",
          "updated",
          "message_count"
        ],
-        "title": "SessionInfo",
+        "title": "SavedWorkstreamInfo",
        "type": "object"
      },
      "HealthResponse": {
@@ -10,9 +10,14 @@ import type {
  ConsoleCreateWsRequest,
  ConsoleCreateWsResponse,
  ConsoleHealthResponse,
+  CreateScheduleRequest,
+  ListScheduleRunsResponse,
+  ListSchedulesResponse,
  NodeDetailResponse,
  NodesOptions,
+  ScheduleInfo,
  StatusResponse,
+  UpdateScheduleRequest,
  WorkstreamsOptions,
 } from "./types.js";

@@ -111,4 +116,40 @@ export class TurnstoneConsole extends BaseClient {
  async health(): Promise<ConsoleHealthResponse> {
    return this.request("GET", "/health");
  }
+
+  // -- Schedules ------------------------------------------------------------
+
+  async listSchedules(): Promise<ListSchedulesResponse> {
+    return this.request("GET", "/v1/api/admin/schedules");
+  }
+
+  async createSchedule(opts: CreateScheduleRequest): Promise<ScheduleInfo> {
+    return this.request("POST", "/v1/api/admin/schedules", { json: opts });
+  }
+
+  async getSchedule(taskId: string): Promise<ScheduleInfo> {
+    return this.request("GET", `/v1/api/admin/schedules/${taskId}`);
+  }
+
+  async updateSchedule(
+    taskId: string,
+    opts: UpdateScheduleRequest,
+  ): Promise<ScheduleInfo> {
+    return this.request("PUT", `/v1/api/admin/schedules/${taskId}`, {
+      json: opts,
+    });
+  }
+
+  async deleteSchedule(taskId: string): Promise<StatusResponse> {
+    return this.request("DELETE", `/v1/api/admin/schedules/${taskId}`);
+  }
+
+  async listScheduleRuns(
+    taskId: string,
+    opts?: { limit?: number },
+  ): Promise<ListScheduleRunsResponse> {
+    return this.request("GET", `/v1/api/admin/schedules/${taskId}/runs`, {
+      params: { limit: opts?.limit ?? 50 },
+    });
+  }
 }
@@ -83,8 +83,8 @@ export type {
  DashboardWorkstream,
  DashboardAggregate,
  DashboardResponse,
-  SessionInfo,
-  ListSessionsResponse,
+  SavedWorkstreamInfo,
+  ListSavedWorkstreamsResponse,
  BackendStatus,
  WorkstreamCounts,
  HealthResponse,
@@ -103,6 +103,12 @@ export type {
  ConsoleCreateWsRequest,
  ConsoleCreateWsResponse,
  ConsoleHealthResponse,
+  CreateScheduleRequest,
+  UpdateScheduleRequest,
+  ScheduleInfo,
+  ScheduleRunInfo,
+  ListSchedulesResponse,
+  ListScheduleRunsResponse,
  TurnResult,
  SendAndWaitOptions,
  NodesOptions,
@@ -8,7 +8,7 @@ import type {
  CreateWorkstreamResponse,
  DashboardResponse,
  HealthResponse,
-  ListSessionsResponse,
+  ListSavedWorkstreamsResponse,
  ListWorkstreamsResponse,
  SendAndWaitOptions,
  SendResponse,
@@ -178,10 +178,10 @@ export class TurnstoneServer extends BaseClient {
    return result;
  }

-  // -- Sessions -------------------------------------------------------------
+  // -- Saved workstreams ----------------------------------------------------

-  async listSessions(): Promise<ListSessionsResponse> {
-    return this.request("GET", "/v1/api/sessions");
+  async listSavedWorkstreams(): Promise<ListSavedWorkstreamsResponse> {
+    return this.request("GET", "/v1/api/workstreams/saved");
  }

  // -- Auth -----------------------------------------------------------------
@@ -71,14 +71,13 @@ export interface CreateWorkstreamRequest {
  name?: string;
  model?: string;
  auto_approve?: boolean;
-  resume_session?: string;
+  resume_ws?: string;
 }

 export interface CreateWorkstreamResponse {
  ws_id: string;
  name: string;
  resumed?: boolean;
-  session_id?: string;
  message_count?: number;
 }

@@ -90,7 +89,6 @@ export interface WorkstreamInfo {
  id: string;
  name: string;
  state: string;
-  session_id?: string | null;
 }

 export interface ListWorkstreamsResponse {
@@ -101,7 +99,6 @@ export interface DashboardWorkstream {
  id: string;
  name: string;
  state: string;
-  session_id?: string | null;
  title?: string;
  tokens?: number;
  context_ratio?: number;
@@ -128,11 +125,11 @@ export interface DashboardResponse {
 }

 // ---------------------------------------------------------------------------
-// Server API — Sessions
+// Server API — Saved workstreams
 // ---------------------------------------------------------------------------

-export interface SessionInfo {
-  session_id: string;
+export interface SavedWorkstreamInfo {
+  ws_id: string;
  alias?: string | null;
  title?: string | null;
  created: string;
@@ -140,8 +137,8 @@ export interface SessionInfo {
  message_count: number;
 }

-export interface ListSessionsResponse {
-  sessions: SessionInfo[];
+export interface ListSavedWorkstreamsResponse {
+  workstreams: SavedWorkstreamInfo[];
 }

 // ---------------------------------------------------------------------------
@@ -269,6 +266,77 @@ export interface ConsoleHealthResponse {
  versions: string[];
 }

+// ---------------------------------------------------------------------------
+// Console API — Schedules
+// ---------------------------------------------------------------------------
+
+export interface CreateScheduleRequest {
+  name: string;
+  schedule_type: string;
+  initial_message: string;
+  description?: string;
+  cron_expr?: string;
+  at_time?: string;
+  target_mode?: string;
+  model?: string;
+  auto_approve?: boolean;
+  auto_approve_tools?: string[];
+  enabled?: boolean;
+}
+
+export interface UpdateScheduleRequest {
+  name?: string;
+  description?: string;
+  schedule_type?: string;
+  cron_expr?: string;
+  at_time?: string;
+  target_mode?: string;
+  model?: string;
+  initial_message?: string;
+  auto_approve?: boolean;
+  auto_approve_tools?: string[];
+  enabled?: boolean;
+}
+
+export interface ScheduleInfo {
+  task_id: string;
+  name: string;
+  description: string;
+  schedule_type: string;
+  cron_expr: string;
+  at_time: string;
+  target_mode: string;
+  model: string;
+  initial_message: string;
+  auto_approve: boolean;
+  auto_approve_tools: string[];
+  enabled: boolean;
+  created_by: string;
+  last_run: string | null;
+  next_run: string | null;
+  created: string;
+  updated: string;
+}
+
+export interface ListSchedulesResponse {
+  schedules: ScheduleInfo[];
+}
+
+export interface ScheduleRunInfo {
+  run_id: string;
+  task_id: string;
+  node_id: string;
+  ws_id: string;
+  correlation_id: string;
+  started: string;
+  status: string;
+  error: string;
+}
+
+export interface ListScheduleRunsResponse {
+  runs: ScheduleRunInfo[];
+}
+
 // ---------------------------------------------------------------------------
 // SDK-specific types
 // ---------------------------------------------------------------------------
@@ -719,7 +719,7 @@ class TestServerAuth:
        srv_mod._metrics.model = "test-model"

        mock_session = MagicMock()
-        mock_session.session_id = "test-session-id"
+        mock_session.ws_id = "test-session-id"

        mock_ws = MagicMock()
        mock_ws.id = "test-ws"
@@ -937,7 +937,7 @@ class TestServerLogin:
        srv_mod._metrics.model = "test-model"

        mock_session = MagicMock()
-        mock_session.session_id = "test-session-id"
+        mock_session.ws_id = "test-session-id"

        mock_ws = MagicMock()
        mock_ws.id = "test-ws"
@@ -534,7 +534,7 @@ class TestWorkstreamModelParam:
            nonlocal captured_alias
            captured_alias = model_alias
            mock_session = MagicMock()
-            mock_session.session_id = "test123"
+            mock_session.ws_id = "test123"
            return mock_session

        mgr = WorkstreamManager(factory)
@@ -548,7 +548,7 @@ class TestWorkstreamModelParam:
            nonlocal captured_alias
            captured_alias = model_alias
            mock_session = MagicMock()
-            mock_session.session_id = "test123"
+            mock_session.ws_id = "test123"
            return mock_session

        from turnstone.core.workstream import WorkstreamManager
@@ -0,0 +1,338 @@
+"""Tests for the channel gateway HTTP notify endpoint."""
+
+from __future__ import annotations
+
+from unittest.mock import AsyncMock
+
+import pytest
+from starlette.testclient import TestClient
+
+from turnstone.channels._http import create_channel_app
+from turnstone.core.storage._sqlite import SQLiteBackend
+
+
+@pytest.fixture
+def storage(tmp_path):
+    return SQLiteBackend(str(tmp_path / "test.db"))
+
+
+@pytest.fixture
+def mock_adapter():
+    adapter = AsyncMock()
+    adapter.channel_type = "discord"
+    adapter.send = AsyncMock(return_value="msg_001")
+    return adapter
+
+
+@pytest.fixture
+def no_auth_client(storage, mock_adapter):
+    """Client with no auth configured (for fail-closed tests)."""
+    app = create_channel_app({"discord": mock_adapter}, storage)
+    return TestClient(app)
+
+
+@pytest.fixture
+def client(storage, mock_adapter):
+    """Default client with static auth token configured."""
+    app = create_channel_app({"discord": mock_adapter}, storage, auth_token="test-secret-token")
+    return TestClient(app)
+
+
+@pytest.fixture
+def authed_client(storage, mock_adapter):
+    """Alias — same as client, for auth-specific test clarity."""
+    app = create_channel_app({"discord": mock_adapter}, storage, auth_token="test-secret-token")
+    return TestClient(app)
+
+
+@pytest.fixture
+def jwt_client(storage, mock_adapter):
+    """Client with JWT auth configured."""
+    app = create_channel_app({"discord": mock_adapter}, storage, jwt_secret="a" * 32)
+    return TestClient(app)
+
+
+class TestNotifyEndpoint:
+    def test_health(self, client):
+        resp = client.get("/health")
+        assert resp.status_code == 200
+        assert resp.json()["status"] == "ok"
+
+    def _headers(self) -> dict[str, str]:
+        return {"Authorization": "Bearer test-secret-token"}
+
+    def test_direct_discord_target(self, client, mock_adapter):
+        resp = client.post(
+            "/v1/api/notify",
+            json={
+                "target": {"channel_type": "discord", "channel_id": "123456"},
+                "message": "Hello!",
+            },
+            headers=self._headers(),
+        )
+        assert resp.status_code == 200
+        results = resp.json()["results"]
+        assert len(results) == 1
+        assert results[0]["status"] == "sent"
+        assert results[0]["message_id"] == "msg_001"
+        mock_adapter.send.assert_called_once_with("123456", "Hello!")
+
+    def test_with_title(self, client, mock_adapter):
+        resp = client.post(
+            "/v1/api/notify",
+            json={
+                "target": {"channel_type": "discord", "channel_id": "123456"},
+                "message": "Hello!",
+                "title": "Alert",
+            },
+            headers=self._headers(),
+        )
+        assert resp.status_code == 200
+        mock_adapter.send.assert_called_once_with("123456", "**Alert**\nHello!")
+
+    def test_username_resolution(self, client, storage, mock_adapter):
+        # Create a user and link a channel
+        storage.create_user("u1", "testuser", "Test User", "hash")
+        storage.create_channel_user("discord", "disc_123", "u1")
+
+        resp = client.post(
+            "/v1/api/notify",
+            json={
+                "target": {"username": "testuser"},
+                "message": "Hello!",
+            },
+            headers=self._headers(),
+        )
+        assert resp.status_code == 200
+        results = resp.json()["results"]
+        assert len(results) == 1
+        assert results[0]["status"] == "sent"
+        mock_adapter.send.assert_called_once_with("disc_123", "Hello!")
+
+    def test_unknown_username(self, client):
+        resp = client.post(
+            "/v1/api/notify",
+            json={
+                "target": {"username": "nobody"},
+                "message": "Hello!",
+            },
+            headers=self._headers(),
+        )
+        assert resp.status_code == 404
+        error = resp.json()["error"]
+        assert "nobody" not in error
+        assert "not found or has no linked channels" in error
+
+    def test_user_no_channels(self, authed_client, storage):
+        storage.create_user("u1", "testuser", "Test User", "hash")
+
+        resp = authed_client.post(
+            "/v1/api/notify",
+            json={
+                "target": {"username": "testuser"},
+                "message": "Hello!",
+            },
+            headers={"Authorization": "Bearer test-secret-token"},
+        )
+        assert resp.status_code == 404
+        # Generic message — must not differentiate "not found" vs "no channels"
+        error = resp.json()["error"]
+        assert "testuser" not in error
+        assert "not found or has no linked channels" in error
+
+    def test_missing_fields(self, client):
+        resp = client.post(
+            "/v1/api/notify",
+            json={"target": {"username": "x"}},
+            headers=self._headers(),
+        )
+        assert resp.status_code == 400
+
+    def test_missing_target(self, client):
+        resp = client.post(
+            "/v1/api/notify",
+            json={"message": "Hello!"},
+            headers=self._headers(),
+        )
+        assert resp.status_code == 400
+
+    def test_invalid_target(self, client):
+        resp = client.post(
+            "/v1/api/notify",
+            json={
+                "target": {"invalid": "field"},
+                "message": "Hello!",
+            },
+            headers=self._headers(),
+        )
+        assert resp.status_code == 400
+
+    def test_no_adapter(self, client, storage):
+        # App has discord adapter, try email target
+        resp = client.post(
+            "/v1/api/notify",
+            json={
+                "target": {"channel_type": "email", "channel_id": "test@example.com"},
+                "message": "Hello!",
+            },
+            headers=self._headers(),
+        )
+        assert resp.status_code == 200
+        results = resp.json()["results"]
+        assert results[0]["status"] == "no_adapter"
+
+    def test_adapter_failure(self, client, mock_adapter):
+        mock_adapter.send.side_effect = RuntimeError("Discord API error")
+        resp = client.post(
+            "/v1/api/notify",
+            json={
+                "target": {"channel_type": "discord", "channel_id": "123456"},
+                "message": "Hello!",
+            },
+            headers=self._headers(),
+        )
+        assert resp.status_code == 200
+        results = resp.json()["results"]
+        assert results[0]["status"] == "failed"
+
+    def test_invalid_json(self, client):
+        resp = client.post(
+            "/v1/api/notify",
+            content=b"not json",
+            headers={
+                "content-type": "application/json",
+                "Authorization": "Bearer test-secret-token",
+            },
+        )
+        assert resp.status_code == 400
+
+    def test_whitespace_only_message(self, client):
+        """Whitespace-only messages should be rejected."""
+        resp = client.post(
+            "/v1/api/notify",
+            json={
+                "target": {"channel_type": "discord", "channel_id": "123"},
+                "message": "   ",
+            },
+            headers=self._headers(),
+        )
+        assert resp.status_code == 400
+
+
+class TestNotifyAuth:
+    """Tests for authentication on the /v1/api/notify endpoint."""
+
+    def test_reject_when_unconfigured(self, no_auth_client):
+        """Requests are rejected (fail closed) when no auth is configured."""
+        resp = no_auth_client.post(
+            "/v1/api/notify",
+            json={
+                "target": {"channel_type": "discord", "channel_id": "123"},
+                "message": "Hello!",
+            },
+        )
+        assert resp.status_code == 401
+
+    def test_reject_without_token(self, authed_client):
+        """Requests without Authorization header are rejected when auth is configured."""
+        resp = authed_client.post(
+            "/v1/api/notify",
+            json={
+                "target": {"channel_type": "discord", "channel_id": "123"},
+                "message": "Hello!",
+            },
+        )
+        assert resp.status_code == 401
+
+    def test_reject_wrong_token(self, authed_client):
+        """Requests with wrong token are rejected."""
+        resp = authed_client.post(
+            "/v1/api/notify",
+            json={
+                "target": {"channel_type": "discord", "channel_id": "123"},
+                "message": "Hello!",
+            },
+            headers={"Authorization": "Bearer wrong-token"},
+        )
+        assert resp.status_code == 401
+
+    def test_accept_valid_static_token(self, authed_client, mock_adapter):
+        """Requests with correct static token are accepted."""
+        resp = authed_client.post(
+            "/v1/api/notify",
+            json={
+                "target": {"channel_type": "discord", "channel_id": "123"},
+                "message": "Hello!",
+            },
+            headers={"Authorization": "Bearer test-secret-token"},
+        )
+        assert resp.status_code == 200
+        assert resp.json()["results"][0]["status"] == "sent"
+
+    def test_accept_valid_jwt(self, jwt_client, mock_adapter):
+        """Requests with a valid JWT for the channel audience are accepted."""
+        from turnstone.core.auth import JWT_AUD_CHANNEL, create_jwt
+
+        token = create_jwt(
+            user_id="system",
+            scopes=frozenset({"write"}),
+            source="service",
+            secret="a" * 32,
+            audience=JWT_AUD_CHANNEL,
+        )
+        resp = jwt_client.post(
+            "/v1/api/notify",
+            json={
+                "target": {"channel_type": "discord", "channel_id": "123"},
+                "message": "Hello!",
+            },
+            headers={"Authorization": f"Bearer {token}"},
+        )
+        assert resp.status_code == 200
+
+    def test_reject_jwt_wrong_audience(self, jwt_client):
+        """JWTs with wrong audience are rejected."""
+        from turnstone.core.auth import create_jwt
+
+        token = create_jwt(
+            user_id="system",
+            scopes=frozenset({"write"}),
+            source="service",
+            secret="a" * 32,
+            audience="turnstone-server",  # wrong audience
+        )
+        resp = jwt_client.post(
+            "/v1/api/notify",
+            json={
+                "target": {"channel_type": "discord", "channel_id": "123"},
+                "message": "Hello!",
+            },
+            headers={"Authorization": f"Bearer {token}"},
+        )
+        assert resp.status_code == 401
+
+    def test_reject_jwt_wrong_secret(self, jwt_client):
+        """JWTs signed with wrong secret are rejected."""
+        from turnstone.core.auth import JWT_AUD_CHANNEL, create_jwt
+
+        token = create_jwt(
+            user_id="system",
+            scopes=frozenset({"write"}),
+            source="service",
+            secret="b" * 32,  # wrong secret
+            audience=JWT_AUD_CHANNEL,
+        )
+        resp = jwt_client.post(
+            "/v1/api/notify",
+            json={
+                "target": {"channel_type": "discord", "channel_id": "123"},
+                "message": "Hello!",
+            },
+            headers={"Authorization": f"Bearer {token}"},
+        )
+        assert resp.status_code == 401
+
+    def test_health_bypasses_auth(self, authed_client):
+        """Health endpoint is always accessible regardless of auth config."""
+        resp = authed_client.get("/health")
+        assert resp.status_code == 200
@@ -0,0 +1,618 @@
+"""Tests for the notify tool (prepare + execute) in ChatSession."""
+
+from __future__ import annotations
+
+from typing import TYPE_CHECKING
+from unittest.mock import MagicMock
+
+if TYPE_CHECKING:
+    from turnstone.core.session import ChatSession
+
+
+def _make_session() -> ChatSession:
+    """Create a minimal ChatSession with mocked dependencies."""
+    from unittest.mock import patch
+
+    with (
+        patch("turnstone.core.memory.register_workstream"),
+        patch("turnstone.core.session.save_message"),
+    ):
+        from turnstone.core.session import ChatSession
+
+        ui = MagicMock()
+        session = ChatSession(
+            client=MagicMock(),
+            model="test-model",
+            ui=ui,
+            instructions=None,
+            temperature=0.7,
+            max_tokens=1000,
+            tool_timeout=30,
+        )
+    return session
+
+
+class TestPrepareNotify:
+    def test_valid_username_target(self):
+        session = _make_session()
+        result = session._prepare_notify(
+            "call_1",
+            {
+                "message": "Hello!",
+                "username": "admin",
+            },
+        )
+        assert "execute" in result
+        assert result["func_name"] == "notify"
+        assert result["needs_approval"] is False
+        assert "@admin" in result["header"]
+        assert result["username"] == "admin"
+        assert result["message"] == "Hello!"
+
+    def test_valid_direct_target(self):
+        session = _make_session()
+        result = session._prepare_notify(
+            "call_1",
+            {
+                "message": "Hello!",
+                "channel_type": "discord",
+                "channel_id": "123456",
+            },
+        )
+        assert "execute" in result
+        assert result["channel_type"] == "discord"
+        assert result["channel_id"] == "123456"
+        assert "discord:123456" in result["header"]
+
+    def test_missing_message(self):
+        session = _make_session()
+        result = session._prepare_notify("call_1", {"username": "admin"})
+        assert "error" in result
+        assert "message" in result["error"].lower()
+
+    def test_empty_message(self):
+        session = _make_session()
+        result = session._prepare_notify(
+            "call_1",
+            {
+                "message": "",
+                "username": "admin",
+            },
+        )
+        assert "error" in result
+
+    def test_message_too_long(self):
+        session = _make_session()
+        result = session._prepare_notify(
+            "call_1",
+            {
+                "message": "x" * 2001,
+                "username": "admin",
+            },
+        )
+        assert "error" in result
+        assert "2000" in result["error"]
+
+    def test_both_username_and_direct(self):
+        session = _make_session()
+        result = session._prepare_notify(
+            "call_1",
+            {
+                "message": "Hello!",
+                "username": "admin",
+                "channel_type": "discord",
+                "channel_id": "123",
+            },
+        )
+        assert "error" in result
+        assert "both" in result["error"].lower() or "ambiguous" in result["error"].lower()
+
+    def test_no_target(self):
+        session = _make_session()
+        result = session._prepare_notify("call_1", {"message": "Hello!"})
+        assert "error" in result
+
+    def test_channel_type_without_id(self):
+        session = _make_session()
+        result = session._prepare_notify(
+            "call_1",
+            {
+                "message": "Hello!",
+                "channel_type": "discord",
+            },
+        )
+        assert "error" in result
+        assert "channel_id" in result["error"]
+
+    def test_channel_id_without_type(self):
+        session = _make_session()
+        result = session._prepare_notify(
+            "call_1",
+            {
+                "message": "Hello!",
+                "channel_id": "123456",
+            },
+        )
+        assert "error" in result
+        assert "channel_type" in result["error"]
+
+    def test_preview_truncated(self):
+        session = _make_session()
+        result = session._prepare_notify(
+            "call_1",
+            {
+                "message": "a" * 200,
+                "username": "admin",
+            },
+        )
+        assert result["preview"].endswith("...")
+        assert len(result["preview"]) <= 123  # 120 chars + "..."
+
+    def test_title_passed_through(self):
+        session = _make_session()
+        result = session._prepare_notify(
+            "call_1",
+            {
+                "message": "Hello!",
+                "username": "admin",
+                "title": "Alert",
+            },
+        )
+        assert result["title"] == "Alert"
+
+
+class TestExecNotify:
+    def test_sends_http_to_channel_gateway(self, tmp_path):
+        from turnstone.core.storage._sqlite import SQLiteBackend
+
+        storage = SQLiteBackend(str(tmp_path / "test.db"))
+        storage.register_service("channel", "ch-1", "http://localhost:8091")
+
+        session = _make_session()
+        item = {
+            "call_id": "call_1",
+            "func_name": "notify",
+            "message": "Hello!",
+            "username": "admin",
+            "channel_type": "",
+            "channel_id": "",
+            "title": "Alert",
+        }
+
+        mock_resp = MagicMock()
+        mock_resp.status_code = 200
+        mock_resp.json.return_value = {
+            "results": [{"channel_type": "discord", "channel_id": "123", "status": "sent"}]
+        }
+
+        from unittest.mock import patch
+
+        with (
+            patch("turnstone.core.session.get_storage", return_value=storage),
+            patch("turnstone.core.session.httpx.post", return_value=mock_resp) as mock_post,
+            patch.dict("os.environ", {}, clear=False),
+        ):
+            call_id, msg = session._exec_notify(item)
+
+        assert call_id == "call_1"
+        assert "sent successfully" in msg.lower()
+        mock_post.assert_called_once()
+        post_kwargs = mock_post.call_args
+        assert post_kwargs.kwargs["json"]["target"] == {"username": "admin"}
+        assert post_kwargs.kwargs["json"]["message"] == "Hello!"
+
+    def test_no_services_available(self, tmp_path):
+        from turnstone.core.storage._sqlite import SQLiteBackend
+
+        storage = SQLiteBackend(str(tmp_path / "test.db"))
+        # No services registered
+
+        session = _make_session()
+        item = {
+            "call_id": "call_1",
+            "func_name": "notify",
+            "message": "Hello!",
+            "username": "admin",
+            "channel_type": "",
+            "channel_id": "",
+            "title": "",
+        }
+
+        from unittest.mock import patch
+
+        with (
+            patch("turnstone.core.session.get_storage", return_value=storage),
+            patch("turnstone.core.session.time.sleep"),
+        ):
+            call_id, msg = session._exec_notify(item)
+
+        assert "no channel gateway" in msg.lower()
+
+    def test_rate_limit(self, tmp_path):
+        from turnstone.core.storage._sqlite import SQLiteBackend
+
+        storage = SQLiteBackend(str(tmp_path / "test.db"))
+        storage.register_service("channel", "ch-1", "http://localhost:8091")
+
+        session = _make_session()
+        item = {
+            "call_id": "call_1",
+            "func_name": "notify",
+            "message": "Hello!",
+            "username": "admin",
+            "channel_type": "",
+            "channel_id": "",
+            "title": "",
+        }
+
+        mock_resp = MagicMock()
+        mock_resp.status_code = 200
+        mock_resp.json.return_value = {
+            "results": [{"channel_type": "discord", "channel_id": "123", "status": "sent"}]
+        }
+
+        from unittest.mock import patch
+
+        with (
+            patch("turnstone.core.session.get_storage", return_value=storage),
+            patch("turnstone.core.session.httpx.post", return_value=mock_resp),
+            patch.dict("os.environ", {}, clear=False),
+        ):
+            for _i in range(5):
+                call_id, msg = session._exec_notify(item)
+                assert "sent successfully" in msg.lower()
+
+            # 6th should fail
+            call_id, msg = session._exec_notify(item)
+            assert "rate limit" in msg.lower()
+
+    def test_rate_limit_not_consumed_on_failure(self, tmp_path):
+        """Failed delivery should not consume rate limit slots."""
+        from turnstone.core.storage._sqlite import SQLiteBackend
+
+        storage = SQLiteBackend(str(tmp_path / "test.db"))
+        storage.register_service("channel", "ch-1", "http://localhost:8091")
+
+        session = _make_session()
+        item = {
+            "call_id": "call_1",
+            "func_name": "notify",
+            "message": "Hello!",
+            "username": "admin",
+            "channel_type": "",
+            "channel_id": "",
+            "title": "",
+        }
+
+        from unittest.mock import patch
+
+        with (
+            patch("turnstone.core.session.get_storage", return_value=storage),
+            patch(
+                "turnstone.core.session.httpx.post",
+                side_effect=ConnectionError("refused"),
+            ),
+            patch.dict("os.environ", {}, clear=False),
+            patch("turnstone.core.session.time.sleep"),
+        ):
+            # All fail — counter should stay at 0
+            for _i in range(3):
+                session._exec_notify(item)
+            assert session._notify_count == 0
+
+    def test_counter_on_init(self):
+        session = _make_session()
+        assert session._notify_count == 0
+
+    def test_http_failure_reported(self, tmp_path):
+        from turnstone.core.storage._sqlite import SQLiteBackend
+
+        storage = SQLiteBackend(str(tmp_path / "test.db"))
+        storage.register_service("channel", "ch-1", "http://localhost:8091")
+
+        session = _make_session()
+        item = {
+            "call_id": "call_1",
+            "func_name": "notify",
+            "message": "Hello!",
+            "username": "",
+            "channel_type": "discord",
+            "channel_id": "999",
+            "title": "",
+        }
+
+        from unittest.mock import patch
+
+        with (
+            patch("turnstone.core.session.get_storage", return_value=storage),
+            patch(
+                "turnstone.core.session.httpx.post",
+                side_effect=ConnectionError("refused"),
+            ),
+            patch.dict("os.environ", {}, clear=False),
+            patch("turnstone.core.session.time.sleep"),
+        ):
+            call_id, msg = session._exec_notify(item)
+
+        # Error message should be generic (no internal details)
+        assert "delivery failed" in msg.lower()
+        assert "refused" not in msg
+        assert "ch-1" not in msg
+
+    def test_first_healthy_only(self, tmp_path):
+        """Only the first healthy gateway should receive the request."""
+        from turnstone.core.storage._sqlite import SQLiteBackend
+
+        storage = SQLiteBackend(str(tmp_path / "test.db"))
+        storage.register_service("channel", "ch-1", "http://localhost:8091")
+        storage.register_service("channel", "ch-2", "http://localhost:8092")
+
+        session = _make_session()
+        item = {
+            "call_id": "call_1",
+            "func_name": "notify",
+            "message": "Hello!",
+            "username": "admin",
+            "channel_type": "",
+            "channel_id": "",
+            "title": "",
+        }
+
+        mock_resp = MagicMock()
+        mock_resp.status_code = 200
+        mock_resp.json.return_value = {
+            "results": [{"channel_type": "discord", "channel_id": "123", "status": "sent"}]
+        }
+
+        from unittest.mock import patch
+
+        with (
+            patch("turnstone.core.session.get_storage", return_value=storage),
+            patch("turnstone.core.session.httpx.post", return_value=mock_resp) as mock_post,
+            patch.dict("os.environ", {}, clear=False),
+        ):
+            session._exec_notify(item)
+
+        # Should only have been called once (first healthy)
+        assert mock_post.call_count == 1
+
+    def test_ssrf_protection(self, tmp_path):
+        """URLs with non-http(s) schemes should be skipped."""
+        from turnstone.core.storage._sqlite import SQLiteBackend
+
+        storage = SQLiteBackend(str(tmp_path / "test.db"))
+        # Register a service with an invalid scheme
+        storage.register_service("channel", "ch-bad", "ftp://evil.example.com")
+
+        session = _make_session()
+        item = {
+            "call_id": "call_1",
+            "func_name": "notify",
+            "message": "Hello!",
+            "username": "admin",
+            "channel_type": "",
+            "channel_id": "",
+            "title": "",
+        }
+
+        from unittest.mock import patch
+
+        with (
+            patch("turnstone.core.session.get_storage", return_value=storage),
+            patch("turnstone.core.session.httpx.post") as mock_post,
+            patch.dict("os.environ", {}, clear=False),
+            patch("turnstone.core.session.time.sleep"),
+        ):
+            call_id, msg = session._exec_notify(item)
+
+        # httpx.post should never be called for ftp:// URL
+        mock_post.assert_not_called()
+        assert "delivery failed" in msg.lower()
+
+    def test_retry_on_no_services(self, tmp_path):
+        """Retries service lookup when no gateways are initially available."""
+        session = _make_session()
+        item = {
+            "call_id": "call_1",
+            "func_name": "notify",
+            "message": "Hello!",
+            "username": "admin",
+            "channel_type": "",
+            "channel_id": "",
+            "title": "",
+        }
+
+        # First two calls return empty, third returns a service
+        call_count = 0
+
+        def _list_services(stype: str, max_age_seconds: int = 120) -> list[dict[str, str]]:
+            nonlocal call_count
+            call_count += 1
+            if call_count <= 2:
+                return []
+            return [
+                {
+                    "service_type": "channel",
+                    "service_id": "ch-1",
+                    "url": "http://localhost:8091",
+                    "metadata": "{}",
+                    "last_heartbeat": "",
+                    "created": "",
+                }
+            ]
+
+        mock_resp = MagicMock()
+        mock_resp.status_code = 200
+        mock_resp.json.return_value = {
+            "results": [{"channel_type": "discord", "channel_id": "123", "status": "sent"}]
+        }
+
+        from unittest.mock import patch
+
+        mock_storage = MagicMock()
+        mock_storage.list_services = _list_services
+
+        with (
+            patch("turnstone.core.session.get_storage", return_value=mock_storage),
+            patch("turnstone.core.session.httpx.post", return_value=mock_resp),
+            patch.dict("os.environ", {}, clear=False),
+            patch("turnstone.core.session.time.sleep") as mock_sleep,
+        ):
+            call_id, msg = session._exec_notify(item)
+
+        assert "sent successfully" in msg.lower()
+        # Should have slept twice (retry delays)
+        assert mock_sleep.call_count == 2
+
+    def test_retry_on_all_gateways_failed(self, tmp_path):
+        """Retries when all gateways fail on first attempt but succeed on retry."""
+        from turnstone.core.storage._sqlite import SQLiteBackend
+
+        storage = SQLiteBackend(str(tmp_path / "test.db"))
+        storage.register_service("channel", "ch-1", "http://localhost:8091")
+
+        session = _make_session()
+        item = {
+            "call_id": "call_1",
+            "func_name": "notify",
+            "message": "Hello!",
+            "username": "admin",
+            "channel_type": "",
+            "channel_id": "",
+            "title": "",
+        }
+
+        call_count = 0
+
+        def _post(*args, **kwargs):
+            nonlocal call_count
+            call_count += 1
+            if call_count <= 1:
+                raise ConnectionError("refused")
+            resp = MagicMock()
+            resp.status_code = 200
+            resp.json.return_value = {
+                "results": [{"channel_type": "discord", "channel_id": "123", "status": "sent"}]
+            }
+            return resp
+
+        from unittest.mock import patch
+
+        with (
+            patch("turnstone.core.session.get_storage", return_value=storage),
+            patch("turnstone.core.session.httpx.post", side_effect=_post),
+            patch.dict("os.environ", {}, clear=False),
+            patch("turnstone.core.session.time.sleep") as mock_sleep,
+        ):
+            call_id, msg = session._exec_notify(item)
+
+        assert "sent successfully" in msg.lower()
+        assert mock_sleep.call_count == 1
+
+    def test_no_services_logs_warning(self, tmp_path):
+        """Server-side warning is logged when no services are available."""
+        from turnstone.core.storage._sqlite import SQLiteBackend
+
+        storage = SQLiteBackend(str(tmp_path / "test.db"))
+
+        session = _make_session()
+        item = {
+            "call_id": "call_1",
+            "func_name": "notify",
+            "message": "Hello!",
+            "username": "admin",
+            "channel_type": "",
+            "channel_id": "",
+            "title": "",
+        }
+
+        from unittest.mock import patch
+
+        with (
+            patch("turnstone.core.session.get_storage", return_value=storage),
+            patch("turnstone.core.session.time.sleep"),
+            patch("turnstone.core.session.log") as mock_log,
+        ):
+            session._exec_notify(item)
+
+        # Should have logged warnings for retries + final exhaustion
+        warning_calls = [c for c in mock_log.warning.call_args_list]
+        assert len(warning_calls) >= 3  # 2 retry warnings + 1 exhaustion
+        events = [c.args[0] for c in warning_calls]
+        assert "notify.no_services" in events
+        assert "notify.no_services_exhausted" in events
+
+    def test_all_gateways_failed_logs_warning(self, tmp_path):
+        """Server-side warning is logged when all gateways fail."""
+        from turnstone.core.storage._sqlite import SQLiteBackend
+
+        storage = SQLiteBackend(str(tmp_path / "test.db"))
+        storage.register_service("channel", "ch-1", "http://localhost:8091")
+
+        session = _make_session()
+        item = {
+            "call_id": "call_1",
+            "func_name": "notify",
+            "message": "Hello!",
+            "username": "admin",
+            "channel_type": "",
+            "channel_id": "",
+            "title": "",
+        }
+
+        from unittest.mock import patch
+
+        with (
+            patch("turnstone.core.session.get_storage", return_value=storage),
+            patch(
+                "turnstone.core.session.httpx.post",
+                side_effect=ConnectionError("refused"),
+            ),
+            patch.dict("os.environ", {}, clear=False),
+            patch("turnstone.core.session.time.sleep"),
+            patch("turnstone.core.session.log") as mock_log,
+        ):
+            session._exec_notify(item)
+
+        warning_calls = [c for c in mock_log.warning.call_args_list]
+        events = [c.args[0] for c in warning_calls]
+        # 2 retry warnings + 1 final failure
+        assert "notify.all_gateways_failed" in events
+        assert "notify.delivery_failed" in events
+
+    def test_gateway_200_but_no_delivery(self, tmp_path):
+        """HTTP 200 with all results failed should not count as success."""
+        from turnstone.core.storage._sqlite import SQLiteBackend
+
+        storage = SQLiteBackend(str(tmp_path / "test.db"))
+        storage.register_service("channel", "ch-1", "http://localhost:8091")
+
+        session = _make_session()
+        item = {
+            "call_id": "call_1",
+            "func_name": "notify",
+            "message": "Hello!",
+            "username": "admin",
+            "channel_type": "",
+            "channel_id": "",
+            "title": "",
+        }
+
+        mock_resp = MagicMock()
+        mock_resp.status_code = 200
+        mock_resp.json.return_value = {
+            "results": [{"channel_type": "discord", "channel_id": "123", "status": "no_adapter"}]
+        }
+
+        from unittest.mock import patch
+
+        with (
+            patch("turnstone.core.session.get_storage", return_value=storage),
+            patch("turnstone.core.session.httpx.post", return_value=mock_resp),
+            patch.dict("os.environ", {}, clear=False),
+            patch("turnstone.core.session.time.sleep"),
+        ):
+            call_id, msg = session._exec_notify(item)
+
+        assert "delivery failed" in msg.lower()
+        assert session._notify_count == 0
@@ -27,7 +27,7 @@ class TestServerSpec:
        expected = {
            "/v1/api/workstreams",
            "/v1/api/dashboard",
-            "/v1/api/sessions",
+            "/v1/api/workstreams/saved",
            "/v1/api/send",
            "/v1/api/approve",
            "/v1/api/plan",
@@ -1103,6 +1103,43 @@ class TestOpenAIParameterGating:
        assert "temperature" not in kwargs
        assert "reasoning_effort" not in kwargs

+    def test_gpt5_pro_unsupported_effort_falls_back(self) -> None:
+        """GPT-5 pro only supports 'high'; unsupported values fall back to default."""
+        caps = self.provider.get_capabilities("gpt-5-pro")
+        kwargs: dict[str, Any] = {}
+        self.provider._apply_model_params(kwargs, caps, temperature=0.7, reasoning_effort="medium")
+        assert "temperature" not in kwargs
+        assert kwargs["reasoning_effort"] == "high"  # fell back to default
+
+    def test_gpt5_pro_supported_effort_passes_through(self) -> None:
+        """GPT-5 pro accepts 'high' directly."""
+        caps = self.provider.get_capabilities("gpt-5-pro")
+        kwargs: dict[str, Any] = {}
+        self.provider._apply_model_params(kwargs, caps, temperature=0.7, reasoning_effort="high")
+        assert kwargs["reasoning_effort"] == "high"
+
+    def test_gpt54_1m_context_and_effort(self) -> None:
+        """GPT-5.4: 1M context, temperature when effort=none, xhigh supported."""
+        caps = self.provider.get_capabilities("gpt-5.4")
+        assert caps.context_window == 1050000
+        kwargs: dict[str, Any] = {}
+        self.provider._apply_model_params(kwargs, caps, temperature=0.7, reasoning_effort="none")
+        assert kwargs["temperature"] == 0.7
+        assert "reasoning_effort" not in kwargs
+        kwargs2: dict[str, Any] = {}
+        self.provider._apply_model_params(kwargs2, caps, temperature=0.7, reasoning_effort="xhigh")
+        assert "temperature" not in kwargs2
+        assert kwargs2["reasoning_effort"] == "xhigh"
+
+    def test_gpt54_pro_no_temperature_always_reasoning(self) -> None:
+        """GPT-5.4 pro: no temperature, medium/high/xhigh only."""
+        caps = self.provider.get_capabilities("gpt-5.4-pro")
+        assert caps.context_window == 1050000
+        kwargs: dict[str, Any] = {}
+        self.provider._apply_model_params(kwargs, caps, temperature=0.7, reasoning_effort="low")
+        assert "temperature" not in kwargs
+        assert kwargs["reasoning_effort"] == "medium"  # fell back from unsupported "low"
+

 class TestAnthropicReasoningNone:
    """Verify 'none' effort disables thinking for manual-thinking models."""
@@ -1,6 +1,6 @@
 """Tests for the atomic workstream resumption flow.

-Covers CreateWorkstreamMessage resume_session field, SessionResumedEvent,
+Covers CreateWorkstreamMessage resume_ws field, WorkstreamResumedEvent,
 WorkstreamCreatedEvent resumed fields, and server endpoint handling.
 """

@@ -10,8 +10,8 @@ import json

 from turnstone.mq.protocol import (
    CreateWorkstreamMessage,
-    SessionResumedEvent,
    WorkstreamCreatedEvent,
+    WorkstreamResumedEvent,
 )

 # ---------------------------------------------------------------------------
@@ -20,93 +20,78 @@ from turnstone.mq.protocol import (


 class TestCreateWorkstreamMessageResumeField:
-    def test_resume_session_defaults_empty(self) -> None:
+    def test_resume_ws_defaults_empty(self) -> None:
        msg = CreateWorkstreamMessage(name="test")
-        assert msg.resume_session == ""
+        assert msg.resume_ws == ""

-    def test_resume_session_set(self) -> None:
-        msg = CreateWorkstreamMessage(name="test", resume_session="sess-abc")
-        assert msg.resume_session == "sess-abc"
+    def test_resume_ws_set(self) -> None:
+        msg = CreateWorkstreamMessage(name="test", resume_ws="ws-abc")
+        assert msg.resume_ws == "ws-abc"

-    def test_resume_session_serializes(self) -> None:
-        msg = CreateWorkstreamMessage(resume_session="sess-xyz")
+    def test_resume_ws_serializes(self) -> None:
+        msg = CreateWorkstreamMessage(resume_ws="ws-xyz")
        data = json.loads(msg.to_json())
-        assert data["resume_session"] == "sess-xyz"
+        assert data["resume_ws"] == "ws-xyz"

-    def test_resume_session_deserializes(self) -> None:
-        msg = CreateWorkstreamMessage(resume_session="sess-123")
+    def test_resume_ws_deserializes(self) -> None:
+        msg = CreateWorkstreamMessage(resume_ws="ws-123")
        raw = msg.to_json()
        from turnstone.mq.protocol import InboundMessage

        restored = InboundMessage.from_json(raw)
-        assert getattr(restored, "resume_session", "") == "sess-123"
+        assert getattr(restored, "resume_ws", "") == "ws-123"


 class TestWorkstreamCreatedEventResumeFields:
    def test_default_not_resumed(self) -> None:
        event = WorkstreamCreatedEvent(ws_id="ws-1", name="test")
        assert event.resumed is False
-        assert event.session_id == ""
        assert event.message_count == 0

    def test_resumed_fields(self) -> None:
-        event = WorkstreamCreatedEvent(
-            ws_id="ws-1", name="test", resumed=True, session_id="s-1", message_count=42
-        )
+        event = WorkstreamCreatedEvent(ws_id="ws-1", name="test", resumed=True, message_count=42)
        assert event.resumed is True
-        assert event.session_id == "s-1"
        assert event.message_count == 42

    def test_serializes_resumed_fields(self) -> None:
-        event = WorkstreamCreatedEvent(
-            ws_id="ws-1", resumed=True, session_id="s-1", message_count=10
-        )
+        event = WorkstreamCreatedEvent(ws_id="ws-1", resumed=True, message_count=10)
        data = json.loads(event.to_json())
        assert data["resumed"] is True
-        assert data["session_id"] == "s-1"
        assert data["message_count"] == 10

    def test_deserializes_resumed_fields(self) -> None:
-        event = WorkstreamCreatedEvent(
-            ws_id="ws-1", resumed=True, session_id="s-1", message_count=5
-        )
+        event = WorkstreamCreatedEvent(ws_id="ws-1", resumed=True, message_count=5)
        from turnstone.mq.protocol import OutboundEvent

        restored = OutboundEvent.from_json(event.to_json())
        assert isinstance(restored, WorkstreamCreatedEvent)
        assert restored.resumed is True
-        assert restored.session_id == "s-1"
        assert restored.message_count == 5


-class TestSessionResumedEvent:
+class TestWorkstreamResumedEvent:
    def test_defaults(self) -> None:
-        event = SessionResumedEvent(ws_id="ws-1")
-        assert event.type == "session_resumed"
-        assert event.session_id == ""
+        event = WorkstreamResumedEvent(ws_id="ws-1")
+        assert event.type == "ws_resumed"
        assert event.message_count == 0
        assert event.name == ""

    def test_with_values(self) -> None:
-        event = SessionResumedEvent(
-            ws_id="ws-1", session_id="s-abc", message_count=25, name="My Chat"
-        )
-        assert event.session_id == "s-abc"
+        event = WorkstreamResumedEvent(ws_id="ws-1", message_count=25, name="My Chat")
        assert event.message_count == 25
        assert event.name == "My Chat"

    def test_round_trip(self) -> None:
-        event = SessionResumedEvent(ws_id="ws-1", session_id="s-abc", message_count=10, name="Chat")
+        event = WorkstreamResumedEvent(ws_id="ws-1", message_count=10, name="Chat")
        from turnstone.mq.protocol import OutboundEvent

        restored = OutboundEvent.from_json(event.to_json())
-        assert isinstance(restored, SessionResumedEvent)
-        assert restored.session_id == "s-abc"
+        assert isinstance(restored, WorkstreamResumedEvent)
        assert restored.message_count == 10
        assert restored.name == "Chat"

    def test_registered_in_outbound_registry(self) -> None:
        from turnstone.mq.protocol import _OUTBOUND_REGISTRY

-        assert "session_resumed" in _OUTBOUND_REGISTRY
-        assert _OUTBOUND_REGISTRY["session_resumed"] is SessionResumedEvent
+        assert "ws_resumed" in _OUTBOUND_REGISTRY
+        assert _OUTBOUND_REGISTRY["ws_resumed"] is WorkstreamResumedEvent
@@ -0,0 +1,264 @@
+"""Tests for scheduled task admin API endpoints."""
+
+from __future__ import annotations
+
+import pytest
+from starlette.applications import Starlette
+from starlette.routing import Mount, Route
+from starlette.testclient import TestClient
+
+from turnstone.console.server import (
+    admin_create_schedule,
+    admin_delete_schedule,
+    admin_get_schedule,
+    admin_list_schedule_runs,
+    admin_list_schedules,
+    admin_update_schedule,
+)
+from turnstone.core.storage._sqlite import SQLiteBackend
+
+
+@pytest.fixture
+def storage(tmp_path):
+    """Fresh SQLite backend for each test."""
+    return SQLiteBackend(str(tmp_path / "test.db"))
+
+
+@pytest.fixture
+def client(storage):
+    """TestClient with storage and auth bypassed."""
+    app = Starlette(
+        routes=[
+            Mount(
+                "/v1",
+                routes=[
+                    Route("/api/admin/schedules", admin_list_schedules),
+                    Route("/api/admin/schedules", admin_create_schedule, methods=["POST"]),
+                    Route("/api/admin/schedules/{task_id}", admin_get_schedule),
+                    Route(
+                        "/api/admin/schedules/{task_id}",
+                        admin_update_schedule,
+                        methods=["PUT"],
+                    ),
+                    Route(
+                        "/api/admin/schedules/{task_id}",
+                        admin_delete_schedule,
+                        methods=["DELETE"],
+                    ),
+                    Route(
+                        "/api/admin/schedules/{task_id}/runs",
+                        admin_list_schedule_runs,
+                    ),
+                ],
+            ),
+        ],
+    )
+    app.state.auth_storage = storage
+    return TestClient(app)
+
+
+def _cron_payload(**overrides):
+    """Build default cron schedule creation payload."""
+    defaults = {
+        "name": "Daily report",
+        "description": "Generate the summary",
+        "schedule_type": "cron",
+        "cron_expr": "0 9 * * *",
+        "target_mode": "auto",
+        "model": "gpt-5",
+        "initial_message": "Generate the daily report",
+    }
+    defaults.update(overrides)
+    return defaults
+
+
+def _at_payload(**overrides):
+    """Build default at-time schedule creation payload."""
+    defaults = {
+        "name": "One-shot task",
+        "description": "Run once",
+        "schedule_type": "at",
+        "at_time": "2099-01-01T00:00:00+00:00",
+        "target_mode": "auto",
+        "model": "gpt-5",
+        "initial_message": "Do the thing",
+    }
+    defaults.update(overrides)
+    return defaults
+
+
+class TestScheduleAPI:
+    """Tests for the 6 admin schedule endpoints."""
+
+    def test_list_empty(self, client):
+        resp = client.get("/v1/api/admin/schedules")
+        assert resp.status_code == 200
+        data = resp.json()
+        assert data["schedules"] == []
+
+    def test_create_cron(self, client):
+        resp = client.post("/v1/api/admin/schedules", json=_cron_payload())
+        assert resp.status_code == 200
+        task = resp.json()
+        assert task["name"] == "Daily report"
+        assert task["schedule_type"] == "cron"
+        assert task["cron_expr"] == "0 9 * * *"
+        assert task["enabled"] is True
+        assert "task_id" in task
+        assert "created" in task
+        assert "next_run" in task
+        assert task["next_run"] != ""
+
+    def test_create_at(self, client):
+        resp = client.post("/v1/api/admin/schedules", json=_at_payload())
+        assert resp.status_code == 200
+        task = resp.json()
+        assert task["schedule_type"] == "at"
+        assert task["at_time"] == "2099-01-01T00:00:00+00:00"
+        assert task["next_run"] == "2099-01-01T00:00:00+00:00"
+
+    def test_create_missing_name(self, client):
+        payload = _cron_payload()
+        del payload["name"]
+        resp = client.post("/v1/api/admin/schedules", json=payload)
+        assert resp.status_code == 400
+        assert "name" in resp.json()["error"].lower()
+
+    def test_create_invalid_cron(self, client):
+        resp = client.post(
+            "/v1/api/admin/schedules",
+            json=_cron_payload(cron_expr="not a cron"),
+        )
+        assert resp.status_code == 400
+        assert "cron" in resp.json()["error"].lower()
+
+    def test_create_naive_at_time(self, client):
+        """Naive timestamps (no timezone) should be rejected."""
+        resp = client.post(
+            "/v1/api/admin/schedules",
+            json=_at_payload(at_time="2099-01-01T00:00:00"),
+        )
+        assert resp.status_code == 400
+        assert "timezone" in resp.json()["error"].lower()
+
+    def test_create_past_at_time(self, client):
+        resp = client.post(
+            "/v1/api/admin/schedules",
+            json=_at_payload(at_time="2000-01-01T00:00:00+00:00"),
+        )
+        assert resp.status_code == 400
+        assert "future" in resp.json()["error"].lower()
+
+    def test_get_schedule(self, client):
+        create_resp = client.post("/v1/api/admin/schedules", json=_cron_payload())
+        task_id = create_resp.json()["task_id"]
+
+        resp = client.get(f"/v1/api/admin/schedules/{task_id}")
+        assert resp.status_code == 200
+        assert resp.json()["task_id"] == task_id
+        assert resp.json()["name"] == "Daily report"
+
+    def test_get_nonexistent(self, client):
+        resp = client.get("/v1/api/admin/schedules/nonexistent_id")
+        assert resp.status_code == 404
+
+    def test_update_schedule(self, client):
+        create_resp = client.post("/v1/api/admin/schedules", json=_cron_payload())
+        task_id = create_resp.json()["task_id"]
+
+        resp = client.put(
+            f"/v1/api/admin/schedules/{task_id}",
+            json={"name": "Weekly report"},
+        )
+        assert resp.status_code == 200
+        assert resp.json()["name"] == "Weekly report"
+
+        # Verify via GET
+        get_resp = client.get(f"/v1/api/admin/schedules/{task_id}")
+        assert get_resp.json()["name"] == "Weekly report"
+
+    def test_update_nonexistent(self, client):
+        resp = client.put(
+            "/v1/api/admin/schedules/nonexistent_id",
+            json={"name": "Nope"},
+        )
+        assert resp.status_code == 404
+
+    def test_delete_schedule(self, client):
+        create_resp = client.post("/v1/api/admin/schedules", json=_cron_payload())
+        task_id = create_resp.json()["task_id"]
+
+        resp = client.delete(f"/v1/api/admin/schedules/{task_id}")
+        assert resp.status_code == 200
+        assert resp.json()["status"] == "ok"
+
+        # Verify gone
+        get_resp = client.get(f"/v1/api/admin/schedules/{task_id}")
+        assert get_resp.status_code == 404
+
+    def test_delete_nonexistent(self, client):
+        resp = client.delete("/v1/api/admin/schedules/nonexistent_id")
+        assert resp.status_code == 404
+
+    def test_list_runs_empty(self, client):
+        create_resp = client.post("/v1/api/admin/schedules", json=_cron_payload())
+        task_id = create_resp.json()["task_id"]
+
+        resp = client.get(f"/v1/api/admin/schedules/{task_id}/runs")
+        assert resp.status_code == 200
+        assert resp.json()["runs"] == []
+
+    def test_list_runs_nonexistent(self, client):
+        resp = client.get("/v1/api/admin/schedules/nonexistent_id/runs")
+        assert resp.status_code == 404
+
+    def test_create_specific_node_target(self, client):
+        payload = _cron_payload(target_mode="node-custom-001")
+        resp = client.post("/v1/api/admin/schedules", json=payload)
+        assert resp.status_code == 200
+        data = resp.json()
+        assert data["target_mode"] == "node-custom-001"
+
+    def test_list_runs_with_data(self, client, storage):
+        create_resp = client.post("/v1/api/admin/schedules", json=_cron_payload())
+        task_id = create_resp.json()["task_id"]
+
+        # Record runs directly in storage
+        storage.record_task_run(
+            run_id="run_001",
+            task_id=task_id,
+            node_id="node-1",
+            ws_id="ws_abc",
+            correlation_id="corr_001",
+            started="2025-06-01T09:00:00",
+            status="dispatched",
+            error="",
+        )
+        storage.record_task_run(
+            run_id="run_002",
+            task_id=task_id,
+            node_id="node-2",
+            ws_id="",
+            correlation_id="corr_002",
+            started="2025-06-01T09:01:00",
+            status="failed",
+            error="No reachable nodes",
+        )
+
+        resp = client.get(f"/v1/api/admin/schedules/{task_id}/runs")
+        assert resp.status_code == 200
+        runs = resp.json()["runs"]
+        assert len(runs) == 2
+        # Most recent first
+        assert runs[0]["run_id"] == "run_002"
+        assert runs[0]["status"] == "failed"
+        assert runs[1]["run_id"] == "run_001"
+
+    def test_list_runs_invalid_limit(self, client):
+        create_resp = client.post("/v1/api/admin/schedules", json=_cron_payload())
+        task_id = create_resp.json()["task_id"]
+
+        # Invalid limit should not crash — falls back to 50
+        resp = client.get(f"/v1/api/admin/schedules/{task_id}/runs?limit=abc")
+        assert resp.status_code == 200
+        assert resp.json()["runs"] == []
@@ -0,0 +1,259 @@
+"""Tests for scheduled_tasks and scheduled_task_runs storage CRUD."""
+
+from __future__ import annotations
+
+import time
+
+import pytest
+
+from turnstone.core.storage._sqlite import SQLiteBackend
+
+
+@pytest.fixture
+def db(tmp_path):
+    """Fresh SQLite backend for each test."""
+    backend = SQLiteBackend(str(tmp_path / "test.db"))
+    return backend
+
+
+def _make_task_kwargs(**overrides):
+    """Build default kwargs for create_scheduled_task."""
+    defaults = {
+        "task_id": "task_001",
+        "name": "Daily report",
+        "description": "Generate the daily summary",
+        "schedule_type": "cron",
+        "cron_expr": "0 9 * * *",
+        "at_time": "",
+        "target_mode": "auto",
+        "model": "gpt-5",
+        "initial_message": "Generate the daily report",
+        "auto_approve": False,
+        "auto_approve_tools": [],
+        "created_by": "u_admin",
+        "next_run": "2099-01-01T09:00:00",
+    }
+    defaults.update(overrides)
+    return defaults
+
+
+class TestScheduledTaskCRUD:
+    """Tests for scheduled_tasks table operations."""
+
+    def test_create_and_get(self, db):
+        db.create_scheduled_task(**_make_task_kwargs())
+        result = db.get_scheduled_task("task_001")
+        assert result is not None
+        assert result["task_id"] == "task_001"
+        assert result["name"] == "Daily report"
+        assert result["description"] == "Generate the daily summary"
+        assert result["schedule_type"] == "cron"
+        assert result["cron_expr"] == "0 9 * * *"
+        assert result["at_time"] == ""
+        assert result["target_mode"] == "auto"
+        assert result["model"] == "gpt-5"
+        assert result["initial_message"] == "Generate the daily report"
+        assert result["auto_approve"] == 0
+        assert result["auto_approve_tools"] == ""
+        assert result["enabled"] == 1
+        assert result["created_by"] == "u_admin"
+        assert result["next_run"] == "2099-01-01T09:00:00"
+        assert "created" in result
+        assert "updated" in result
+
+    def test_get_nonexistent(self, db):
+        assert db.get_scheduled_task("no_such_task") is None
+
+    def test_create_duplicate_noop(self, db):
+        db.create_scheduled_task(**_make_task_kwargs(name="First"))
+        db.create_scheduled_task(**_make_task_kwargs(name="Second"))
+        result = db.get_scheduled_task("task_001")
+        assert result is not None
+        assert result["name"] == "First"  # first write wins
+
+    def test_list_tasks(self, db):
+        db.create_scheduled_task(**_make_task_kwargs(task_id="task_a", name="Alpha"))
+        # Ensure different created timestamps (resolution is 1 second)
+        time.sleep(1.1)
+        db.create_scheduled_task(**_make_task_kwargs(task_id="task_b", name="Beta"))
+        tasks = db.list_scheduled_tasks()
+        assert len(tasks) == 2
+        # Ordered by created DESC — most recent first
+        assert tasks[0]["task_id"] == "task_b"
+        assert tasks[1]["task_id"] == "task_a"
+
+    def test_update_task(self, db):
+        db.create_scheduled_task(**_make_task_kwargs())
+        original = db.get_scheduled_task("task_001")
+        assert original is not None
+        original_updated = original["updated"]
+
+        time.sleep(0.05)
+        result = db.update_scheduled_task("task_001", name="Weekly report")
+        assert result is True
+
+        updated = db.get_scheduled_task("task_001")
+        assert updated is not None
+        assert updated["name"] == "Weekly report"
+        assert updated["updated"] >= original_updated
+
+    def test_update_enable_disable(self, db):
+        db.create_scheduled_task(**_make_task_kwargs())
+        task = db.get_scheduled_task("task_001")
+        assert task is not None
+        assert task["enabled"] == 1
+
+        db.update_scheduled_task("task_001", enabled=False)
+        task = db.get_scheduled_task("task_001")
+        assert task is not None
+        assert task["enabled"] == 0
+
+        db.update_scheduled_task("task_001", enabled=True)
+        task = db.get_scheduled_task("task_001")
+        assert task is not None
+        assert task["enabled"] == 1
+
+    def test_delete_task(self, db):
+        db.create_scheduled_task(**_make_task_kwargs())
+        assert db.delete_scheduled_task("task_001") is True
+        assert db.get_scheduled_task("task_001") is None
+        # Deleting again returns False
+        assert db.delete_scheduled_task("task_001") is False
+
+    def test_delete_cascades_runs(self, db):
+        db.create_scheduled_task(**_make_task_kwargs())
+        db.record_task_run(
+            run_id="run_001",
+            task_id="task_001",
+            node_id="node_1",
+            ws_id="ws_abc",
+            correlation_id="corr_001",
+            started="2025-01-01T09:00:00",
+            status="dispatched",
+            error="",
+        )
+        assert len(db.list_task_runs("task_001")) == 1
+
+        db.delete_scheduled_task("task_001")
+        assert db.list_task_runs("task_001") == []
+
+    def test_list_due_tasks(self, db):
+        db.create_scheduled_task(
+            **_make_task_kwargs(task_id="past", next_run="2020-01-01T00:00:00")
+        )
+        db.create_scheduled_task(
+            **_make_task_kwargs(task_id="future", next_run="2099-12-31T23:59:59")
+        )
+        now = "2025-06-01T12:00:00"
+        due = db.list_due_tasks(now)
+        assert len(due) == 1
+        assert due[0]["task_id"] == "past"
+
+    def test_list_due_tasks_skips_disabled(self, db):
+        db.create_scheduled_task(
+            **_make_task_kwargs(task_id="disabled_task", next_run="2020-01-01T00:00:00")
+        )
+        db.update_scheduled_task("disabled_task", enabled=False)
+        due = db.list_due_tasks("2025-06-01T12:00:00")
+        assert len(due) == 0
+
+    def test_list_due_tasks_empty_next_run(self, db):
+        db.create_scheduled_task(**_make_task_kwargs(task_id="empty_next", next_run=""))
+        due = db.list_due_tasks("2099-12-31T23:59:59")
+        assert len(due) == 0
+
+    def test_at_task_fields(self, db):
+        db.create_scheduled_task(
+            **_make_task_kwargs(
+                task_id="at_task",
+                schedule_type="at",
+                cron_expr="",
+                at_time="2099-06-15T14:00:00",
+                next_run="2099-06-15T14:00:00",
+            )
+        )
+        result = db.get_scheduled_task("at_task")
+        assert result is not None
+        assert result["schedule_type"] == "at"
+        assert result["at_time"] == "2099-06-15T14:00:00"
+
+
+class TestScheduledTaskRuns:
+    """Tests for scheduled_task_runs table operations."""
+
+    def test_record_and_list(self, db):
+        db.create_scheduled_task(**_make_task_kwargs())
+        db.record_task_run(
+            run_id="run_a",
+            task_id="task_001",
+            node_id="node_1",
+            ws_id="ws_1",
+            correlation_id="corr_a",
+            started="2025-01-01T09:00:00",
+            status="dispatched",
+            error="",
+        )
+        db.record_task_run(
+            run_id="run_b",
+            task_id="task_001",
+            node_id="node_2",
+            ws_id="ws_2",
+            correlation_id="corr_b",
+            started="2025-01-02T09:00:00",
+            status="dispatched",
+            error="",
+        )
+        runs = db.list_task_runs("task_001")
+        assert len(runs) == 2
+        # Ordered by started DESC — most recent first
+        assert runs[0]["run_id"] == "run_b"
+        assert runs[1]["run_id"] == "run_a"
+
+    def test_list_runs_respects_limit(self, db):
+        db.create_scheduled_task(**_make_task_kwargs())
+        for i in range(3):
+            db.record_task_run(
+                run_id=f"run_{i}",
+                task_id="task_001",
+                node_id="node_1",
+                ws_id="",
+                correlation_id=f"corr_{i}",
+                started=f"2025-01-0{i + 1}T09:00:00",
+                status="dispatched",
+                error="",
+            )
+        runs = db.list_task_runs("task_001", limit=2)
+        assert len(runs) == 2
+
+    def test_list_runs_empty(self, db):
+        assert db.list_task_runs("no_such_task") == []
+
+    def test_prune_task_runs(self, db):
+        db.create_scheduled_task(**_make_task_kwargs())
+        # Old run (should be pruned)
+        db.record_task_run(
+            run_id="old_run",
+            task_id="task_001",
+            node_id="node_1",
+            ws_id="",
+            correlation_id="c_old",
+            started="2020-01-01T00:00:00",
+            status="dispatched",
+            error="",
+        )
+        # Recent run (should survive)
+        db.record_task_run(
+            run_id="new_run",
+            task_id="task_001",
+            node_id="node_1",
+            ws_id="",
+            correlation_id="c_new",
+            started="2099-01-01T00:00:00",
+            status="dispatched",
+            error="",
+        )
+        pruned = db.prune_task_runs(retention_days=90)
+        assert pruned == 1
+        runs = db.list_task_runs("task_001")
+        assert len(runs) == 1
+        assert runs[0]["run_id"] == "new_run"
@@ -0,0 +1,272 @@
+"""Tests for turnstone.console.scheduler — TaskScheduler tick and dispatch."""
+
+from __future__ import annotations
+
+from unittest.mock import MagicMock
+
+import pytest
+
+from turnstone.console.scheduler import TaskScheduler
+
+
+@pytest.fixture
+def mocks():
+    """Broker, collector, and storage mocks for scheduler tests."""
+    broker = MagicMock()
+    broker._redis = MagicMock()
+    collector = MagicMock()
+    storage = MagicMock()
+    return broker, collector, storage
+
+
+def _make_task(**overrides):
+    """Build a minimal task dict matching storage row format."""
+    defaults = {
+        "task_id": "task_001",
+        "name": "Test task",
+        "description": "",
+        "schedule_type": "cron",
+        "cron_expr": "0 9 * * *",
+        "at_time": "",
+        "target_mode": "auto",
+        "model": "gpt-5",
+        "initial_message": "Run the tests",
+        "auto_approve": 0,
+        "auto_approve_tools": "",
+        "enabled": 1,
+        "created_by": "u_admin",
+        "next_run": "2020-01-01T09:00:00",
+        "last_run": "",
+        "created": "2020-01-01T00:00:00",
+        "updated": "2020-01-01T00:00:00",
+    }
+    defaults.update(overrides)
+    return defaults
+
+
+def _make_node(node_id="node-001", reachable=True, ws_total=2, max_ws=10):
+    """Build a minimal node dict matching collector output."""
+    return {
+        "node_id": node_id,
+        "reachable": reachable,
+        "ws_total": ws_total,
+        "max_ws": max_ws,
+    }
+
+
+class TestSchedulerTick:
+    """Tests for _tick() lock acquisition and dispatch logic."""
+
+    def test_tick_acquires_lock(self, mocks):
+        broker, collector, storage = mocks
+        broker._redis.set.return_value = True
+        storage.list_due_tasks.return_value = []
+
+        scheduler = TaskScheduler(broker, collector, storage)
+        scheduler._tick()
+
+        broker._redis.set.assert_called_once()
+        storage.list_due_tasks.assert_called_once()
+        # Lock released via Lua eval (conditional delete)
+        broker._redis.eval.assert_called_once()
+
+    def test_tick_skips_when_locked(self, mocks):
+        broker, collector, storage = mocks
+        broker._redis.set.return_value = None  # lock held by another console
+
+        scheduler = TaskScheduler(broker, collector, storage)
+        scheduler._tick()
+
+        storage.list_due_tasks.assert_not_called()
+
+    def test_dispatch_auto_mode(self, mocks):
+        broker, collector, storage = mocks
+        broker._redis.set.return_value = True
+
+        task = _make_task(target_mode="auto")
+        storage.list_due_tasks.return_value = [task]
+        collector.get_nodes.return_value = ([_make_node()], 1)
+
+        scheduler = TaskScheduler(broker, collector, storage)
+        scheduler._tick()
+
+        broker.push_inbound.assert_called_once()
+        _, kwargs = broker.push_inbound.call_args
+        assert (
+            kwargs.get("node_id") == "node-001"
+            or broker.push_inbound.call_args[1].get("node_id") == "node-001"
+        )
+        storage.record_task_run.assert_called_once()
+        run_kwargs = storage.record_task_run.call_args[1]
+        assert run_kwargs["node_id"] == "node-001"
+        assert run_kwargs["status"] == "dispatched"
+
+    def test_dispatch_pool_mode(self, mocks):
+        broker, collector, storage = mocks
+        broker._redis.set.return_value = True
+
+        task = _make_task(target_mode="pool")
+        storage.list_due_tasks.return_value = [task]
+
+        scheduler = TaskScheduler(broker, collector, storage)
+        scheduler._tick()
+
+        broker.push_inbound.assert_called_once()
+        # Pool dispatch calls push_inbound without node_id kwarg
+        args, kwargs = broker.push_inbound.call_args
+        assert kwargs.get("node_id") is None or "node_id" not in kwargs
+        storage.record_task_run.assert_called_once()
+        run_kwargs = storage.record_task_run.call_args[1]
+        assert run_kwargs["node_id"] == "pool"
+
+    def test_dispatch_all_mode(self, mocks):
+        broker, collector, storage = mocks
+        broker._redis.set.return_value = True
+
+        task = _make_task(target_mode="all")
+        storage.list_due_tasks.return_value = [task]
+        collector.get_nodes.return_value = (
+            [_make_node("node-001"), _make_node("node-002")],
+            2,
+        )
+
+        scheduler = TaskScheduler(broker, collector, storage)
+        scheduler._tick()
+
+        assert broker.push_inbound.call_count == 2
+        assert storage.record_task_run.call_count == 2
+
+    def test_dispatch_specific_node(self, mocks):
+        broker, collector, storage = mocks
+        broker._redis.set.return_value = True
+
+        task = _make_task(target_mode="node-001")
+        storage.list_due_tasks.return_value = [task]
+
+        scheduler = TaskScheduler(broker, collector, storage)
+        scheduler._tick()
+
+        broker.push_inbound.assert_called_once()
+        _, kwargs = broker.push_inbound.call_args
+        assert kwargs["node_id"] == "node-001"
+
+    def test_at_task_disables_after_dispatch(self, mocks):
+        broker, collector, storage = mocks
+        broker._redis.set.return_value = True
+
+        task = _make_task(schedule_type="at", cron_expr="", at_time="2099-01-01T00:00:00")
+        storage.list_due_tasks.return_value = [task]
+        collector.get_nodes.return_value = ([_make_node()], 1)
+
+        scheduler = TaskScheduler(broker, collector, storage)
+        scheduler._tick()
+
+        # At-task should be disabled after dispatch
+        update_calls = storage.update_scheduled_task.call_args_list
+        assert len(update_calls) == 1
+        args, kwargs = update_calls[0]
+        assert args[0] == "task_001"
+        assert kwargs["enabled"] is False
+        assert kwargs["next_run"] == ""
+
+    def test_cron_task_updates_next_run(self, mocks):
+        broker, collector, storage = mocks
+        broker._redis.set.return_value = True
+
+        task = _make_task(schedule_type="cron", cron_expr="0 9 * * *")
+        storage.list_due_tasks.return_value = [task]
+        collector.get_nodes.return_value = ([_make_node()], 1)
+
+        scheduler = TaskScheduler(broker, collector, storage)
+        scheduler._tick()
+
+        update_calls = storage.update_scheduled_task.call_args_list
+        assert len(update_calls) == 1
+        _, kwargs = update_calls[0]
+        assert kwargs["next_run"] != ""
+        assert "enabled" not in kwargs  # cron tasks stay enabled
+
+    def test_no_reachable_nodes_records_failure(self, mocks):
+        broker, collector, storage = mocks
+        broker._redis.set.return_value = True
+
+        task = _make_task(target_mode="auto")
+        storage.list_due_tasks.return_value = [task]
+        # No reachable nodes
+        collector.get_nodes.return_value = (
+            [_make_node("node-001", reachable=False)],
+            1,
+        )
+
+        scheduler = TaskScheduler(broker, collector, storage)
+        scheduler._tick()
+
+        broker.push_inbound.assert_not_called()
+        storage.record_task_run.assert_called_once()
+        run_kwargs = storage.record_task_run.call_args[1]
+        assert run_kwargs["status"] == "failed"
+        assert run_kwargs["error"] != ""
+
+    def test_failure_does_not_advance_schedule(self, mocks):
+        """When dispatch fails, last_run/next_run should not be updated."""
+        broker, collector, storage = mocks
+        broker._redis.set.return_value = True
+
+        task = _make_task(target_mode="auto")
+        storage.list_due_tasks.return_value = [task]
+        collector.get_nodes.return_value = ([], 0)  # no nodes at all
+
+        scheduler = TaskScheduler(broker, collector, storage)
+        scheduler._tick()
+
+        # update_scheduled_task should NOT be called (no last_run/next_run advance)
+        storage.update_scheduled_task.assert_not_called()
+
+    def test_fan_out_capped(self, mocks):
+        """Fan-out 'all' mode should respect max_fan_out limit."""
+        broker, collector, storage = mocks
+        broker._redis.set.return_value = True
+
+        task = _make_task(target_mode="all")
+        storage.list_due_tasks.return_value = [task]
+        # 10 reachable nodes but max_fan_out=3
+        nodes = [_make_node(f"node-{i:03d}") for i in range(10)]
+        collector.get_nodes.return_value = (nodes, 10)
+
+        scheduler = TaskScheduler(broker, collector, storage, max_fan_out=3)
+        scheduler._tick()
+
+        assert broker.push_inbound.call_count == 3
+        assert storage.record_task_run.call_count == 3
+
+    def test_specific_node_target(self, mocks):
+        """Non-enum target_mode is treated as a specific node_id."""
+        broker, collector, storage = mocks
+        broker._redis.set.return_value = True
+
+        task = _make_task(target_mode="node-custom-123")
+        storage.list_due_tasks.return_value = [task]
+
+        scheduler = TaskScheduler(broker, collector, storage)
+        scheduler._tick()
+
+        broker.push_inbound.assert_called_once()
+        call_kwargs = broker.push_inbound.call_args
+        assert call_kwargs[1]["node_id"] == "node-custom-123"
+
+    def test_user_id_in_dispatched_message(self, mocks):
+        """Dispatched message should include created_by as user_id."""
+        import json
+
+        broker, collector, storage = mocks
+        broker._redis.set.return_value = True
+
+        task = _make_task(target_mode="pool", created_by="u_scheduler_admin")
+        storage.list_due_tasks.return_value = [task]
+
+        scheduler = TaskScheduler(broker, collector, storage)
+        scheduler._tick()
+
+        msg_json = broker.push_inbound.call_args[0][0]
+        msg_data = json.loads(msg_json)
+        assert msg_data["user_id"] == "u_scheduler_admin"
@@ -2,6 +2,8 @@

 from __future__ import annotations

+import json
+
 import httpx
 import pytest

@@ -240,3 +242,140 @@ async def test_query_params_passed():
        assert "state=running" in captured_url[0]
        assert "page=2" in captured_url[0]
        assert "per_page=25" in captured_url[0]
+
+
+# ---------------------------------------------------------------------------
+# Schedules
+# ---------------------------------------------------------------------------
+
+_SCHEDULE_FIXTURE = {
+    "task_id": "t1",
+    "name": "nightly",
+    "description": "",
+    "schedule_type": "cron",
+    "cron_expr": "0 2 * * *",
+    "at_time": "",
+    "target_mode": "auto",
+    "model": "",
+    "initial_message": "Run nightly checks",
+    "auto_approve": False,
+    "auto_approve_tools": [],
+    "enabled": True,
+    "created_by": "u1",
+    "last_run": None,
+    "next_run": "2026-03-06T02:00:00Z",
+    "created": "2026-03-05T12:00:00Z",
+    "updated": "2026-03-05T12:00:00Z",
+}
+
+
+@pytest.mark.anyio
+async def test_list_schedules():
+    transport = _mock_transport(
+        {"GET /v1/api/admin/schedules": _json_response({"schedules": [_SCHEDULE_FIXTURE]})}
+    )
+    async with httpx.AsyncClient(transport=transport, base_url="http://test") as hc:
+        client = AsyncTurnstoneConsole(httpx_client=hc)
+        resp = await client.list_schedules()
+        assert len(resp.schedules) == 1
+        assert resp.schedules[0].task_id == "t1"
+        assert resp.schedules[0].name == "nightly"
+
+
+@pytest.mark.anyio
+async def test_create_schedule():
+    captured_body: list[dict] = []
+
+    def handler(request: httpx.Request) -> httpx.Response:
+        captured_body.append(json.loads(request.content))
+        return _json_response(_SCHEDULE_FIXTURE)
+
+    transport = httpx.MockTransport(handler)
+    async with httpx.AsyncClient(transport=transport, base_url="http://test") as hc:
+        client = AsyncTurnstoneConsole(httpx_client=hc)
+        resp = await client.create_schedule(
+            name="nightly",
+            schedule_type="cron",
+            initial_message="Run nightly checks",
+            cron_expr="0 2 * * *",
+        )
+        assert resp.task_id == "t1"
+        body = captured_body[0]
+        assert body["name"] == "nightly"
+        assert body["schedule_type"] == "cron"
+        assert body["cron_expr"] == "0 2 * * *"
+        assert body["initial_message"] == "Run nightly checks"
+        # Optional fields with defaults should not appear when not set
+        assert "description" not in body
+        assert "model" not in body
+
+
+@pytest.mark.anyio
+async def test_get_schedule():
+    transport = _mock_transport(
+        {"GET /v1/api/admin/schedules/t1": _json_response(_SCHEDULE_FIXTURE)}
+    )
+    async with httpx.AsyncClient(transport=transport, base_url="http://test") as hc:
+        client = AsyncTurnstoneConsole(httpx_client=hc)
+        resp = await client.get_schedule("t1")
+        assert resp.task_id == "t1"
+        assert resp.schedule_type == "cron"
+
+
+@pytest.mark.anyio
+async def test_update_schedule_partial():
+    """Only explicitly-passed fields should appear in the request body."""
+    captured_body: list[dict] = []
+
+    def handler(request: httpx.Request) -> httpx.Response:
+        captured_body.append(json.loads(request.content))
+        return _json_response({**_SCHEDULE_FIXTURE, "enabled": False})
+
+    transport = httpx.MockTransport(handler)
+    async with httpx.AsyncClient(transport=transport, base_url="http://test") as hc:
+        client = AsyncTurnstoneConsole(httpx_client=hc)
+        resp = await client.update_schedule("t1", enabled=False)
+        assert resp.enabled is False
+        body = captured_body[0]
+        assert body == {"enabled": False}
+
+
+@pytest.mark.anyio
+async def test_delete_schedule():
+    transport = _mock_transport(
+        {"DELETE /v1/api/admin/schedules/t1": _json_response({"status": "ok"})}
+    )
+    async with httpx.AsyncClient(transport=transport, base_url="http://test") as hc:
+        client = AsyncTurnstoneConsole(httpx_client=hc)
+        resp = await client.delete_schedule("t1")
+        assert resp.status == "ok"
+
+
+@pytest.mark.anyio
+async def test_list_schedule_runs():
+    transport = _mock_transport(
+        {
+            "GET /v1/api/admin/schedules/t1/runs": _json_response(
+                {
+                    "runs": [
+                        {
+                            "run_id": "r1",
+                            "task_id": "t1",
+                            "node_id": "n1",
+                            "ws_id": "ws1",
+                            "correlation_id": "c1",
+                            "started": "2026-03-05T02:00:00Z",
+                            "status": "dispatched",
+                            "error": "",
+                        }
+                    ]
+                }
+            )
+        }
+    )
+    async with httpx.AsyncClient(transport=transport, base_url="http://test") as hc:
+        client = AsyncTurnstoneConsole(httpx_client=hc)
+        resp = await client.list_schedule_runs("t1", limit=10)
+        assert len(resp.runs) == 1
+        assert resp.runs[0].run_id == "r1"
+        assert resp.runs[0].status == "dispatched"
@@ -148,19 +148,19 @@ async def test_command():


 # ---------------------------------------------------------------------------
-# Sessions
+# History
 # ---------------------------------------------------------------------------


@pytest.mark.anyio
-async def test_list_sessions():
+async def test_list_saved_workstreams():
    transport = _mock_transport(
        {
-            "GET /v1/api/sessions": _json_response(
+            "GET /v1/api/workstreams/saved": _json_response(
                {
-                    "sessions": [
+                    "workstreams": [
                        {
-                            "session_id": "s1",
+                            "ws_id": "s1",
                            "title": "test",
                            "created": "2024-01-01",
                            "updated": "2024-01-02",
@@ -173,8 +173,8 @@ async def test_list_sessions():
    )
    async with httpx.AsyncClient(transport=transport, base_url="http://test") as hc:
        client = AsyncTurnstoneServer(httpx_client=hc)
-        resp = await client.list_sessions()
-        assert len(resp.sessions) == 1
+        resp = await client.list_saved_workstreams()
+        assert len(resp.workstreams) == 1


 # ---------------------------------------------------------------------------
@@ -609,7 +609,7 @@ class TestServerHealthMetrics:
        mock_ui._ws_context_ratio = 0.0

        mock_session = MagicMock()
-        mock_session.session_id = "test-session-id"
+        mock_session.ws_id = "test-session-id"

        mock_ws = MagicMock()
        mock_ws.id = "test-ws"
@@ -785,7 +785,7 @@ class TestServerRateLimiting:
        mock_ui._ws_context_ratio = 0.0

        mock_session = MagicMock()
-        mock_session.session_id = "test-session-id"
+        mock_session.ws_id = "test-session-id"

        mock_ws = MagicMock()
        mock_ws.id = "test-ws"
@@ -0,0 +1,86 @@
+"""Tests for the services registry storage methods."""
+
+from __future__ import annotations
+
+import pytest
+
+from turnstone.core.storage._sqlite import SQLiteBackend
+
+
+@pytest.fixture
+def storage(tmp_path):
+    return SQLiteBackend(str(tmp_path / "test.db"))
+
+
+class TestServiceRegistry:
+    def test_register_and_list(self, storage):
+        storage.register_service("channel", "ch-1", "http://localhost:8091")
+        services = storage.list_services("channel", max_age_seconds=120)
+        assert len(services) == 1
+        assert services[0]["service_type"] == "channel"
+        assert services[0]["service_id"] == "ch-1"
+        assert services[0]["url"] == "http://localhost:8091"
+
+    def test_register_upsert(self, storage):
+        storage.register_service("channel", "ch-1", "http://old:8091")
+        storage.register_service("channel", "ch-1", "http://new:8091")
+        services = storage.list_services("channel", max_age_seconds=120)
+        assert len(services) == 1
+        assert services[0]["url"] == "http://new:8091"
+
+    def test_heartbeat(self, storage):
+        storage.register_service("channel", "ch-1", "http://localhost:8091")
+        result = storage.heartbeat_service("channel", "ch-1")
+        assert result is True
+
+    def test_heartbeat_nonexistent(self, storage):
+        result = storage.heartbeat_service("channel", "nonexistent")
+        assert result is False
+
+    def test_list_filters_stale(self, storage):
+        storage.register_service("channel", "ch-1", "http://localhost:8091")
+        # Manually set heartbeat to the past so it's stale
+        from datetime import UTC, datetime, timedelta
+
+        import sqlalchemy as sa
+
+        from turnstone.core.storage._schema import services
+
+        old_time = (datetime.now(UTC) - timedelta(seconds=300)).strftime("%Y-%m-%dT%H:%M:%S")
+        with storage._engine.connect() as conn:
+            conn.execute(sa.update(services).values(last_heartbeat=old_time))
+            conn.commit()
+
+        # Should be excluded with 120s max age
+        result = storage.list_services("channel", max_age_seconds=120)
+        assert len(result) == 0
+
+    def test_list_empty(self, storage):
+        services = storage.list_services("channel", max_age_seconds=120)
+        assert services == []
+
+    def test_list_filters_by_type(self, storage):
+        storage.register_service("channel", "ch-1", "http://localhost:8091")
+        storage.register_service("bridge", "br-1", "http://localhost:8080")
+        channels = storage.list_services("channel", max_age_seconds=120)
+        bridges = storage.list_services("bridge", max_age_seconds=120)
+        assert len(channels) == 1
+        assert len(bridges) == 1
+
+    def test_deregister(self, storage):
+        storage.register_service("channel", "ch-1", "http://localhost:8091")
+        result = storage.deregister_service("channel", "ch-1")
+        assert result is True
+        services = storage.list_services("channel", max_age_seconds=120)
+        assert services == []
+
+    def test_deregister_nonexistent(self, storage):
+        result = storage.deregister_service("channel", "nonexistent")
+        assert result is False
+
+    def test_metadata(self, storage):
+        storage.register_service(
+            "channel", "ch-1", "http://localhost:8091", metadata='{"adapter": "discord"}'
+        )
+        services = storage.list_services("channel", max_age_seconds=120)
+        assert services[0]["metadata"] == '{"adapter": "discord"}'
@@ -161,12 +161,12 @@ class TestPlanExec:

        return call_id, content, captured.get("messages", [])

-    def test_plan_file_uses_session_id(self, tmp_db, tmp_path, monkeypatch):
-        """Plan file is named .plan-<session_id>.md, not .plan.md."""
+    def test_plan_file_uses_ws_id(self, tmp_db, tmp_path, monkeypatch):
+        """Plan file is named .plan-<ws_id>.md, not .plan.md."""
        monkeypatch.chdir(tmp_path)
        session = _make_session()
        self._run_plan(session, "add feature")
-        expected = tmp_path / f".plan-{session._session_id}.md"
+        expected = tmp_path / f".plan-{session._ws_id}.md"
        assert expected.exists(), f"Expected {expected} to be created"
        assert not (tmp_path / ".plan.md").exists()

@@ -176,7 +176,7 @@ class TestPlanExec:
        session = _make_session()
        plan_content = "## Goal\n\nAdd a new endpoint."
        self._run_plan(session, "add endpoint", agent_return=plan_content)
-        plan_file = tmp_path / f".plan-{session._session_id}.md"
+        plan_file = tmp_path / f".plan-{session._ws_id}.md"
        assert plan_file.read_text() == plan_content

    def test_two_sessions_produce_different_files(self, tmp_db, tmp_path, monkeypatch):
@@ -184,7 +184,7 @@ class TestPlanExec:
        monkeypatch.chdir(tmp_path)
        s1 = _make_session()
        s2 = _make_session()
-        assert s1._session_id != s2._session_id
+        assert s1._ws_id != s2._ws_id
        self._run_plan(s1, "feature A")
        self._run_plan(s2, "feature B")
        files = list(tmp_path.glob(".plan-*.md"))
@@ -1,155 +1,155 @@
-"""Tests for session persistence and resume functionality."""
+"""Tests for workstream persistence and resume functionality."""

 from unittest.mock import MagicMock

 import sqlalchemy as sa

 from turnstone.core.memory import (
-    delete_session,
-    list_sessions,
-    load_session_config,
-    load_session_messages,
-    prune_sessions,
-    register_session,
-    resolve_session,
+    delete_workstream,
+    list_workstreams_with_history,
+    load_messages,
+    load_workstream_config,
+    prune_workstreams,
+    register_workstream,
+    resolve_workstream,
    save_message,
-    save_session_config,
-    set_session_alias,
-    update_session_title,
+    save_workstream_config,
+    set_workstream_alias,
+    update_workstream_title,
 )
 from turnstone.core.session import ChatSession
 from turnstone.core.storage import get_storage

-# ── Session registration ──────────────────────────────────────────────
+# ── Workstream registration ───────────────────────────────────────────


-class TestRegisterSession:
+class TestRegisterWorkstream:
    def test_register_creates_row(self, tmp_db):
-        register_session("abc123")
-        # Session exists in DB (resolve works) even without messages
-        assert resolve_session("abc123") == "abc123"
+        register_workstream("abc123")
+        # Workstream exists in DB (resolve works) even without messages
+        assert resolve_workstream("abc123") == "abc123"

    def test_register_with_title(self, tmp_db):
-        register_session("abc123", title="My Session")
+        register_workstream("abc123", name="My Workstream")
        save_message("abc123", "user", "hello")
-        rows = list_sessions()
-        assert rows[0][2] == "My Session"  # title
+        rows = list_workstreams_with_history()
+        assert rows[0][2] is None  # title column (name is separate)

    def test_register_idempotent(self, tmp_db):
-        register_session("abc123", title="First")
-        register_session("abc123", title="Second")  # should be ignored
+        register_workstream("abc123")
+        update_workstream_title("abc123", "First")
+        register_workstream("abc123")  # should be ignored
+        update_workstream_title("abc123", "First")  # title is set via update
        save_message("abc123", "user", "hello")
-        rows = list_sessions()
+        rows = list_workstreams_with_history()
        assert len(rows) == 1
-        assert rows[0][2] == "First"  # original title preserved
+        assert rows[0][2] == "First"  # title preserved

    def test_update_title(self, tmp_db):
-        register_session("abc123")
-        update_session_title("abc123", "New Title")
+        register_workstream("abc123")
+        update_workstream_title("abc123", "New Title")
        save_message("abc123", "user", "hello")
-        rows = list_sessions()
+        rows = list_workstreams_with_history()
        assert rows[0][2] == "New Title"


-# ── Session alias ─────────────────────────────────────────────────────
+# ── Workstream alias ──────────────────────────────────────────────────


-class TestSessionAlias:
+class TestWorkstreamAlias:
    def test_set_alias(self, tmp_db):
-        register_session("abc123")
-        assert set_session_alias("abc123", "my-session") is True
+        register_workstream("abc123")
+        assert set_workstream_alias("abc123", "my-session") is True
        save_message("abc123", "user", "hello")
-        rows = list_sessions()
+        rows = list_workstreams_with_history()
        assert rows[0][1] == "my-session"  # alias

    def test_alias_conflict(self, tmp_db):
-        register_session("abc123")
-        register_session("def456")
-        set_session_alias("abc123", "taken")
-        assert set_session_alias("def456", "taken") is False
+        register_workstream("abc123")
+        register_workstream("def456")
+        set_workstream_alias("abc123", "taken")
+        assert set_workstream_alias("def456", "taken") is False

-    def test_alias_same_session_ok(self, tmp_db):
-        register_session("abc123")
-        set_session_alias("abc123", "mine")
-        assert set_session_alias("abc123", "mine") is True  # no-op, same session
+    def test_alias_same_workstream_ok(self, tmp_db):
+        register_workstream("abc123")
+        set_workstream_alias("abc123", "mine")
+        assert set_workstream_alias("abc123", "mine") is True  # no-op, same workstream


-# ── Session resolution ────────────────────────────────────────────────
+# ── Workstream resolution ─────────────────────────────────────────────


-class TestResolveSession:
+class TestResolveWorkstream:
    def test_resolve_by_alias(self, tmp_db):
-        register_session("abc123")
-        set_session_alias("abc123", "my-alias")
-        assert resolve_session("my-alias") == "abc123"
+        register_workstream("abc123")
+        set_workstream_alias("abc123", "my-alias")
+        assert resolve_workstream("my-alias") == "abc123"

    def test_resolve_by_exact_id(self, tmp_db):
-        register_session("abc123def456")
-        assert resolve_session("abc123def456") == "abc123def456"
+        register_workstream("abc123def456")
+        assert resolve_workstream("abc123def456") == "abc123def456"

    def test_resolve_by_prefix(self, tmp_db):
-        register_session("abc123def456")
-        assert resolve_session("abc123") == "abc123def456"
+        register_workstream("abc123def456")
+        assert resolve_workstream("abc123") == "abc123def456"

    def test_resolve_prefix_ambiguous(self, tmp_db):
-        register_session("abc123aaaaaa")
-        register_session("abc123bbbbbb")
+        register_workstream("abc123aaaaaa")
+        register_workstream("abc123bbbbbb")
        # Ambiguous prefix should return None
-        assert resolve_session("abc123") is None
+        assert resolve_workstream("abc123") is None

    def test_resolve_not_found(self, tmp_db):
-        assert resolve_session("nonexistent") is None
-
-    def test_resolve_legacy_session(self, tmp_db):
-        """Sessions that exist only in conversations (pre-migration) should auto-register."""
-        save_message("legacy123456", "user", "old message")
-        result = resolve_session("legacy123456")
-        assert result == "legacy123456"
-        # Should now appear in sessions list
-        rows = list_sessions()
-        assert any(r[0] == "legacy123456" for r in rows)
+        assert resolve_workstream("nonexistent") is None


-# ── List sessions ─────────────────────────────────────────────────────
+# ── List workstreams with history ──────────────────────────────────────


-class TestListSessions:
+class TestListWorkstreamsWithHistory:
    def test_empty(self, tmp_db):
-        assert list_sessions() == []
+        assert list_workstreams_with_history() == []

    def test_ordered_by_updated(self, tmp_db):
-        register_session("first")
+        register_workstream("first")
        save_message("first", "user", "hello")
-        register_session("second")
+        # Force an older timestamp so ordering is deterministic
+        engine = get_storage()._engine  # noqa: SLF001
+        with engine.connect() as conn:
+            conn.execute(
+                sa.text("UPDATE workstreams SET updated = '2020-01-01' WHERE ws_id = 'first'")
+            )
+            conn.commit()
+        register_workstream("second")
        save_message("second", "user", "hello")
        # second is more recent
-        rows = list_sessions()
+        rows = list_workstreams_with_history()
        assert rows[0][0] == "second"
        assert rows[1][0] == "first"

    def test_includes_message_count(self, tmp_db):
-        register_session("sess1")
+        register_workstream("sess1")
        save_message("sess1", "user", "hello")
        save_message("sess1", "assistant", "hi")
-        rows = list_sessions()
+        rows = list_workstreams_with_history()
        assert rows[0][5] == 2  # msg_count

    def test_respects_limit(self, tmp_db):
        for i in range(5):
-            register_session(f"sess{i}")
+            register_workstream(f"sess{i}")
            save_message(f"sess{i}", "user", "hello")
-        rows = list_sessions(limit=3)
+        rows = list_workstreams_with_history(limit=3)
        assert len(rows) == 3


-# ── Load session messages ─────────────────────────────────────────────
+# ── Load messages ─────────────────────────────────────────────────────


-class TestLoadSessionMessages:
+class TestLoadMessages:
    def test_simple_user_assistant(self, tmp_db):
        save_message("s1", "user", "hello")
        save_message("s1", "assistant", "hi there")
-        msgs = load_session_messages("s1")
+        msgs = load_messages("s1")
        assert len(msgs) == 2
        assert msgs[0] == {"role": "user", "content": "hello"}
        assert msgs[1] == {"role": "assistant", "content": "hi there"}
@@ -159,7 +159,7 @@ class TestLoadSessionMessages:
        save_message("s1", "assistant", "Let me check.")
        save_message("s1", "tool_call", None, "bash", '{"command":"ls"}', tool_call_id="call_abc")
        save_message("s1", "tool_result", "file1.txt\nfile2.txt", "bash", tool_call_id="call_abc")
-        msgs = load_session_messages("s1")
+        msgs = load_messages("s1")
        assert len(msgs) == 3  # user, assistant+tool_calls, tool
        # Assistant should have content merged with tool_calls
        assert msgs[1]["role"] == "assistant"
@@ -177,7 +177,7 @@ class TestLoadSessionMessages:
        save_message("s1", "user", "do stuff")
        save_message("s1", "tool_call", None, "bash", '{"command":"ls"}')
        save_message("s1", "tool_result", "output", "bash")
-        msgs = load_session_messages("s1")
+        msgs = load_messages("s1")
        assert len(msgs) == 3
        # Synthetic IDs should match
        tc_id = msgs[1]["tool_calls"][0]["id"]
@@ -189,36 +189,36 @@ class TestLoadSessionMessages:
        save_message("s1", "tool_call", None, "search", '{"query":"b"}', tool_call_id="call_2")
        save_message("s1", "tool_result", "result a", "search", tool_call_id="call_1")
        save_message("s1", "tool_result", "result b", "search", tool_call_id="call_2")
-        msgs = load_session_messages("s1")
+        msgs = load_messages("s1")
        assert len(msgs) == 4  # user, assistant+2 tool_calls, 2 tool results
        assert len(msgs[1]["tool_calls"]) == 2
        assert msgs[2]["tool_call_id"] == "call_1"
        assert msgs[3]["tool_call_id"] == "call_2"

-    def test_empty_session(self, tmp_db):
-        assert load_session_messages("nonexistent") == []
+    def test_empty_workstream(self, tmp_db):
+        assert load_messages("nonexistent") == []

    def test_orphaned_tool_result_skipped(self, tmp_db):
        save_message("s1", "user", "hello")
        save_message("s1", "tool_result", "orphan", "bash")
-        msgs = load_session_messages("s1")
+        msgs = load_messages("s1")
        assert len(msgs) == 1  # only the user message


-# ── Delete session ────────────────────────────────────────────────────
+# ── Delete workstream ─────────────────────────────────────────────────


-class TestDeleteSession:
-    def test_delete_removes_session_and_messages(self, tmp_db):
-        register_session("abc123")
+class TestDeleteWorkstream:
+    def test_delete_removes_workstream_and_messages(self, tmp_db):
+        register_workstream("abc123")
        save_message("abc123", "user", "hello")
        save_message("abc123", "assistant", "hi")
-        assert delete_session("abc123") is True
-        assert list_sessions() == []
-        assert load_session_messages("abc123") == []
+        assert delete_workstream("abc123") is True
+        assert list_workstreams_with_history() == []
+        assert load_messages("abc123") == []

    def test_delete_nonexistent(self, tmp_db):
-        assert delete_session("nonexistent") is True  # no-op, still returns True
+        assert delete_workstream("nonexistent") is False


 # ── save_message with tool_call_id ────────────────────────────────────
@@ -230,7 +230,7 @@ class TestSaveMessageToolCallId:
        engine = get_storage()._engine  # noqa: SLF001
        with engine.connect() as conn:
            row = conn.execute(
-                sa.text("SELECT tool_call_id FROM conversations WHERE session_id = 's1'")
+                sa.text("SELECT tool_call_id FROM conversations WHERE ws_id = 's1'")
            ).fetchone()
            assert row[0] == "call_xyz"

@@ -239,20 +239,20 @@ class TestSaveMessageToolCallId:
        engine = get_storage()._engine  # noqa: SLF001
        with engine.connect() as conn:
            row = conn.execute(
-                sa.text("SELECT tool_call_id FROM conversations WHERE session_id = 's1'")
+                sa.text("SELECT tool_call_id FROM conversations WHERE ws_id = 's1'")
            ).fetchone()
            assert row[0] is None


-# ── Sessions table creation ───────────────────────────────────────────
+# ── Workstreams table creation ────────────────────────────────────────


-class TestSessionsTable:
-    def test_sessions_table_exists(self, tmp_db):
+class TestWorkstreamsTable:
+    def test_workstreams_table_exists(self, tmp_db):
        engine = get_storage()._engine  # noqa: SLF001
        with engine.connect() as conn:
            rows = conn.execute(
-                sa.text("SELECT name FROM sqlite_master WHERE type='table' AND name='sessions'")
+                sa.text("SELECT name FROM sqlite_master WHERE type='table' AND name='workstreams'")
            ).fetchall()
            assert len(rows) == 1

@@ -263,15 +263,15 @@ class TestSessionsTable:
            conn.execute(sa.text("SELECT tool_call_id FROM conversations LIMIT 0"))


-# ── ChatSession.resume_session ────────────────────────────────────────
+# ── ChatSession.resume ────────────────────────────────────────────────


-class TestResumeSession:
+class TestResumeWorkstream:
    def test_resume_loads_messages(self, tmp_db, mock_openai_client):
-        # Set up a session with messages in DB
-        register_session("old_sess_123")
-        save_message("old_sess_123", "user", "hello world")
-        save_message("old_sess_123", "assistant", "hi there")
+        # Set up a workstream with messages in DB
+        register_workstream("old_ws_123")
+        save_message("old_ws_123", "user", "hello world")
+        save_message("old_ws_123", "assistant", "hi there")

        # Create a new session and resume
        session = ChatSession(
@@ -283,12 +283,12 @@ class TestResumeSession:
            max_tokens=1000,
            tool_timeout=10,
        )
-        original_id = session._session_id
-        assert original_id != "old_sess_123"
+        original_id = session._ws_id
+        assert original_id != "old_ws_123"

-        result = session.resume_session("old_sess_123")
+        result = session.resume("old_ws_123")
        assert result is True
-        assert session._session_id == "old_sess_123"
+        assert session._ws_id == "old_ws_123"
        assert len(session.messages) == 2
        assert session.messages[0]["content"] == "hello world"
        assert session._title_generated is True
@@ -303,9 +303,9 @@ class TestResumeSession:
            max_tokens=1000,
            tool_timeout=10,
        )
-        assert session.resume_session("nonexistent") is False
+        assert session.resume("nonexistent") is False

-    def test_session_registered_on_init(self, tmp_db, mock_openai_client):
+    def test_workstream_not_registered_until_message(self, tmp_db, mock_openai_client):
        session = ChatSession(
            client=mock_openai_client,
            model="test-model",
@@ -315,20 +315,19 @@ class TestResumeSession:
            max_tokens=1000,
            tool_timeout=10,
        )
-        # Session is registered in DB (resolvable) even before any messages
-        assert resolve_session(session._session_id) == session._session_id
-        # But does not appear in list_sessions until a message is saved
-        assert not any(r[0] == session._session_id for r in list_sessions())
+        # Workstream is not auto-registered on init — only on /new or server creation
+        assert resolve_workstream(session._ws_id) is None
+        assert not any(r[0] == session._ws_id for r in list_workstreams_with_history())


-# ── save_message updates sessions.updated ─────────────────────────────
+# ── save_message updates workstreams.updated ──────────────────────────


-class TestSaveMessageUpdatesSession:
+class TestSaveMessageUpdatesWorkstream:
    def test_updated_timestamp_bumped(self, tmp_db):
-        register_session("s1")
+        register_workstream("s1")
        save_message("s1", "user", "first")
-        rows = list_sessions()
+        rows = list_workstreams_with_history()
        _original_updated = rows[0][4]

        import time
@@ -336,18 +335,18 @@ class TestSaveMessageUpdatesSession:
        time.sleep(0.01)  # ensure different timestamp
        save_message("s1", "user", "hello")

-        rows = list_sessions()
+        rows = list_workstreams_with_history()
        new_updated = rows[0][4]
        # updated should be same or later (sqlite datetime resolution is seconds,
        # so they may be equal in fast tests — just verify no error)
        assert new_updated is not None


-# ── Interrupted session repair ───────────────────────────────────────
+# ── Interrupted workstream repair ─────────────────────────────────────


-class TestInterruptedSessionRepair:
-    """load_session_messages() should strip trailing incomplete tool call turns."""
+class TestInterruptedWorkstreamRepair:
+    """load_messages() should strip trailing incomplete tool call turns."""

    def test_complete_tool_turn_preserved(self, tmp_db):
        """2 tool_calls + 2 tool_results = complete, no stripping."""
@@ -356,7 +355,7 @@ class TestInterruptedSessionRepair:
        save_message("s1", "tool_call", None, "bash", '{"command":"pwd"}', "call_2")
        save_message("s1", "tool_result", "file.txt", tool_call_id="call_1")
        save_message("s1", "tool_result", "/home", tool_call_id="call_2")
-        msgs = load_session_messages("s1")
+        msgs = load_messages("s1")
        assert len(msgs) == 4  # user + assistant(2 calls) + 2 tool results

    def test_partial_tool_results_stripped(self, tmp_db):
@@ -365,7 +364,7 @@ class TestInterruptedSessionRepair:
        save_message("s1", "tool_call", None, "bash", '{"command":"ls"}', "call_1")
        save_message("s1", "tool_call", None, "bash", '{"command":"pwd"}', "call_2")
        save_message("s1", "tool_result", "file.txt", tool_call_id="call_1")
-        msgs = load_session_messages("s1")
+        msgs = load_messages("s1")
        assert len(msgs) == 1  # only user message remains
        assert msgs[0]["role"] == "user"

@@ -375,7 +374,7 @@ class TestInterruptedSessionRepair:
        save_message("s1", "assistant", "Let me check")
        save_message("s1", "tool_call", None, "bash", '{"command":"ls"}', "call_1")
        save_message("s1", "tool_call", None, "bash", '{"command":"pwd"}', "call_2")
-        msgs = load_session_messages("s1")
+        msgs = load_messages("s1")
        # assistant with content was merged into tool_call assistant, so stripped
        assert len(msgs) == 1
        assert msgs[0]["role"] == "user"
@@ -386,42 +385,42 @@ class TestInterruptedSessionRepair:
        save_message("s1", "assistant", "response")
        save_message("s1", "user", "second")
        save_message("s1", "tool_call", None, "bash", '{"command":"ls"}', "call_1")
-        msgs = load_session_messages("s1")
+        msgs = load_messages("s1")
        assert len(msgs) == 3  # user + assistant + user (incomplete turn stripped)
        assert msgs[0]["role"] == "user"
        assert msgs[1]["role"] == "assistant"
        assert msgs[2]["role"] == "user"


-# ── Session config persistence ───────────────────────────────────────
+# ── Workstream config persistence ─────────────────────────────────────


-class TestSessionConfig:
+class TestWorkstreamConfig:
    def test_save_load_roundtrip(self, tmp_db):
        config = {"temperature": "0.3", "reasoning_effort": "high", "creative_mode": "False"}
-        save_session_config("s1", config)
-        loaded = load_session_config("s1")
+        save_workstream_config("s1", config)
+        loaded = load_workstream_config("s1")
        assert loaded == config

    def test_update_existing_key(self, tmp_db):
-        save_session_config("s1", {"temperature": "0.3"})
-        save_session_config("s1", {"temperature": "0.7"})
-        loaded = load_session_config("s1")
+        save_workstream_config("s1", {"temperature": "0.3"})
+        save_workstream_config("s1", {"temperature": "0.7"})
+        loaded = load_workstream_config("s1")
        assert loaded["temperature"] == "0.7"

-    def test_missing_session_returns_empty(self, tmp_db):
-        loaded = load_session_config("nonexistent")
+    def test_missing_workstream_returns_empty(self, tmp_db):
+        loaded = load_workstream_config("nonexistent")
        assert loaded == {}

-    def test_delete_session_removes_config(self, tmp_db):
-        register_session("s1")
+    def test_delete_workstream_removes_config(self, tmp_db):
+        register_workstream("s1")
        save_message("s1", "user", "hi")
-        save_session_config("s1", {"temperature": "0.5"})
-        delete_session("s1")
-        assert load_session_config("s1") == {}
+        save_workstream_config("s1", {"temperature": "0.5"})
+        delete_workstream("s1")
+        assert load_workstream_config("s1") == {}

    def test_resume_restores_config(self, tmp_db):
-        """ChatSession.resume_session() should restore persisted config."""
+        """ChatSession.resume() should restore persisted config."""
        client = MagicMock()
        client.models.list.return_value.data = [MagicMock(id="test-model")]
        ui = MagicMock()
@@ -430,11 +429,11 @@ class TestSessionConfig:
        ui.on_state_change = MagicMock()
        ui.on_rename = MagicMock()

-        # Create a session with specific config
-        register_session("orig")
+        # Create a workstream with specific config
+        register_workstream("orig")
        save_message("orig", "user", "hello")
        save_message("orig", "assistant", "hi there")
-        save_session_config(
+        save_workstream_config(
            "orig",
            {
                "temperature": "0.3",
@@ -456,7 +455,7 @@ class TestSessionConfig:
            tool_timeout=30,
        )
        assert session.temperature == 0.7  # default
-        result = session.resume_session("orig")
+        result = session.resume("orig")
        assert result is True
        assert session.temperature == 0.3
        assert session.reasoning_effort == "high"
@@ -465,84 +464,84 @@ class TestSessionConfig:
        assert session.creative_mode is True


-# ── Prune sessions ───────────────────────────────────────────────────
+# ── Prune workstreams ─────────────────────────────────────────────────


-class TestPruneSessions:
+class TestPruneWorkstreams:
    def test_orphan_removed(self, tmp_db):
-        """Session registered with no messages should be pruned."""
-        register_session("orphan")
-        orphans, stale = prune_sessions()
+        """Workstream registered with no messages should be pruned."""
+        register_workstream("orphan")
+        orphans, stale = prune_workstreams()
        assert orphans == 1
-        assert list_sessions() == []
+        assert list_workstreams_with_history() == []

-    def test_session_with_messages_kept(self, tmp_db):
-        """Session with messages should not be pruned."""
-        register_session("active")
+    def test_workstream_with_messages_kept(self, tmp_db):
+        """Workstream with messages should not be pruned."""
+        register_workstream("active")
        save_message("active", "user", "hello")
-        orphans, _stale = prune_sessions()
+        orphans, _stale = prune_workstreams()
        assert orphans == 0
-        assert len(list_sessions()) == 1
+        assert len(list_workstreams_with_history()) == 1

    def test_stale_unnamed_removed(self, tmp_db):
-        """Old unnamed session should be pruned by retention policy."""
-        register_session("old1")
+        """Old unnamed workstream should be pruned by retention policy."""
+        register_workstream("old1")
        save_message("old1", "user", "ancient message")
        # Force the updated timestamp to the past so it looks stale
        engine = get_storage()._engine  # noqa: SLF001
        with engine.connect() as conn:
            conn.execute(
-                sa.text("UPDATE sessions SET updated = '2020-01-01' WHERE session_id = 'old1'")
+                sa.text("UPDATE workstreams SET updated = '2020-01-01' WHERE ws_id = 'old1'")
            )
            conn.commit()
-        _orphans, stale = prune_sessions(retention_days=30)
+        _orphans, stale = prune_workstreams(retention_days=30)
        assert stale == 1

-    def test_named_session_preserved(self, tmp_db):
-        """Session with alias should be kept regardless of age."""
-        register_session("old2")
-        set_session_alias("old2", "important")
+    def test_named_workstream_preserved(self, tmp_db):
+        """Workstream with alias should be kept regardless of age."""
+        register_workstream("old2")
+        set_workstream_alias("old2", "important")
        save_message("old2", "user", "old but named")
        # Force old timestamp
        engine = get_storage()._engine  # noqa: SLF001
        with engine.connect() as conn:
            conn.execute(
-                sa.text("UPDATE sessions SET updated = '2020-01-01' WHERE session_id = 'old2'")
+                sa.text("UPDATE workstreams SET updated = '2020-01-01' WHERE ws_id = 'old2'")
            )
            conn.commit()
-        _orphans, stale = prune_sessions(retention_days=30)
+        _orphans, stale = prune_workstreams(retention_days=30)
        assert stale == 0
-        assert len(list_sessions()) == 1
+        assert len(list_workstreams_with_history()) == 1

    def test_fresh_unnamed_preserved(self, tmp_db):
-        """Recent unnamed session should not be pruned."""
-        register_session("fresh")
+        """Recent unnamed workstream should not be pruned."""
+        register_workstream("fresh")
        save_message("fresh", "user", "just now")
-        _orphans, stale = prune_sessions(retention_days=30)
+        _orphans, stale = prune_workstreams(retention_days=30)
        assert stale == 0
-        assert len(list_sessions()) == 1
+        assert len(list_workstreams_with_history()) == 1

-    def test_prune_removes_session_config(self, tmp_db):
-        """Pruning orphan/stale sessions should also remove their config rows."""
-        register_session("orphan_cfg")
-        save_session_config("orphan_cfg", {"temperature": "0.5"})
+    def test_prune_removes_workstream_config(self, tmp_db):
+        """Pruning orphan/stale workstreams should also remove their config rows."""
+        register_workstream("orphan_cfg")
+        save_workstream_config("orphan_cfg", {"temperature": "0.5"})

-        register_session("stale_cfg")
+        register_workstream("stale_cfg")
        save_message("stale_cfg", "user", "old")
-        save_session_config("stale_cfg", {"temperature": "0.9"})
+        save_workstream_config("stale_cfg", {"temperature": "0.9"})
        engine = get_storage()._engine  # noqa: SLF001
        with engine.connect() as conn:
            conn.execute(
-                sa.text("UPDATE sessions SET updated = '2020-01-01' WHERE session_id = 'stale_cfg'")
+                sa.text("UPDATE workstreams SET updated = '2020-01-01' WHERE ws_id = 'stale_cfg'")
            )
            conn.commit()

        # Both should have config before prune
-        assert load_session_config("orphan_cfg") == {"temperature": "0.5"}
-        assert load_session_config("stale_cfg") == {"temperature": "0.9"}
+        assert load_workstream_config("orphan_cfg") == {"temperature": "0.5"}
+        assert load_workstream_config("stale_cfg") == {"temperature": "0.9"}

-        prune_sessions(retention_days=30)
+        prune_workstreams(retention_days=30)

        # Config rows should be cleaned up
-        assert load_session_config("orphan_cfg") == {}
-        assert load_session_config("stale_cfg") == {}
+        assert load_workstream_config("orphan_cfg") == {}
+        assert load_workstream_config("stale_cfg") == {}
@@ -14,28 +14,28 @@ def backend(tmp_path):
    reset_storage()


-# -- Session operations --------------------------------------------------------
+# -- Workstream registration ---------------------------------------------------


-class TestRegisterSession:
-    def test_register_creates_session(self, backend):
-        backend.register_session("s1", title="Test")
-        name = backend.get_session_name("s1")
+class TestRegisterWorkstream:
+    def test_register_creates_workstream(self, backend):
+        backend.register_workstream("s1", title="Test")
+        name = backend.get_workstream_display_name("s1")
        assert name == "Test"

    def test_register_idempotent(self, backend):
-        backend.register_session("s1", title="First")
-        backend.register_session("s1", title="Second")
-        name = backend.get_session_name("s1")
+        backend.register_workstream("s1", title="First")
+        backend.register_workstream("s1", title="Second")
+        name = backend.get_workstream_display_name("s1")
        assert name == "First"  # INSERT OR IGNORE preserves first


 class TestSaveAndLoadMessages:
    def test_roundtrip(self, backend):
-        backend.register_session("s1")
+        backend.register_workstream("s1")
        backend.save_message("s1", "user", "hello")
        backend.save_message("s1", "assistant", "world")
-        msgs = backend.load_session_messages("s1")
+        msgs = backend.load_messages("s1")
        assert len(msgs) == 2
        assert msgs[0]["role"] == "user"
        assert msgs[0]["content"] == "hello"
@@ -43,12 +43,12 @@ class TestSaveAndLoadMessages:
        assert msgs[1]["content"] == "world"

    def test_tool_call_grouping(self, backend):
-        backend.register_session("s1")
+        backend.register_workstream("s1")
        backend.save_message("s1", "user", "do something")
        backend.save_message("s1", "tool_call", None, "bash", '{"cmd":"ls"}', tool_call_id="c1")
        backend.save_message("s1", "tool_result", "file.txt", tool_call_id="c1")
        backend.save_message("s1", "assistant", "done")
-        msgs = backend.load_session_messages("s1")
+        msgs = backend.load_messages("s1")
        assert len(msgs) == 4
        assert msgs[1]["role"] == "assistant"
        assert len(msgs[1]["tool_calls"]) == 1
@@ -57,136 +57,136 @@ class TestSaveAndLoadMessages:
        assert msgs[2]["content"] == "file.txt"

    def test_incomplete_turn_repair(self, backend):
-        backend.register_session("s1")
+        backend.register_workstream("s1")
        backend.save_message("s1", "user", "do something")
        backend.save_message("s1", "tool_call", None, "bash", '{"cmd":"ls"}', tool_call_id="c1")
        backend.save_message("s1", "tool_call", None, "read", '{"path":"a"}', tool_call_id="c2")
        # Only 1 result for 2 calls — incomplete turn
        backend.save_message("s1", "tool_result", "ok", tool_call_id="c1")
-        msgs = backend.load_session_messages("s1")
+        msgs = backend.load_messages("s1")
        # Incomplete turn should be stripped
        assert len(msgs) == 1  # only the user message remains

    def test_provider_data_preserved(self, backend):
        import json

-        backend.register_session("s1")
+        backend.register_workstream("s1")
        pd = json.dumps({"encrypted": True})
        backend.save_message("s1", "assistant", "hi", provider_data=pd)
-        msgs = backend.load_session_messages("s1")
+        msgs = backend.load_messages("s1")
        assert msgs[0].get("_provider_content") == {"encrypted": True}

-    def test_empty_session_returns_empty(self, backend):
-        assert backend.load_session_messages("nonexistent") == []
+    def test_empty_workstream_returns_empty(self, backend):
+        assert backend.load_messages("nonexistent") == []


-class TestListSessions:
-    def test_lists_sessions_with_messages(self, backend):
-        backend.register_session("s1")
+class TestListWorkstreamsWithHistory:
+    def test_lists_workstreams_with_messages(self, backend):
+        backend.register_workstream("s1")
        backend.save_message("s1", "user", "hi")
-        backend.register_session("s2")  # no messages
-        rows = backend.list_sessions()
+        backend.register_workstream("s2")  # no messages
+        rows = backend.list_workstreams_with_history()
        assert len(rows) == 1
        assert rows[0][0] == "s1"

    def test_respects_limit(self, backend):
        for i in range(5):
            sid = f"s{i}"
-            backend.register_session(sid)
+            backend.register_workstream(sid)
            backend.save_message(sid, "user", f"msg {i}")
-        rows = backend.list_sessions(limit=3)
+        rows = backend.list_workstreams_with_history(limit=3)
        assert len(rows) == 3


-class TestDeleteSession:
+class TestDeleteWorkstream:
    def test_deletes_all_data(self, backend):
-        backend.register_session("s1")
+        backend.register_workstream("s1")
        backend.save_message("s1", "user", "hi")
-        backend.save_session_config("s1", {"temp": "0.5"})
-        assert backend.delete_session("s1")
-        assert backend.load_session_messages("s1") == []
-        assert backend.load_session_config("s1") == {}
-        assert backend.get_session_name("s1") is None
+        backend.save_workstream_config("s1", {"temp": "0.5"})
+        assert backend.delete_workstream("s1")
+        assert backend.load_messages("s1") == []
+        assert backend.load_workstream_config("s1") == {}
+        assert backend.get_workstream_display_name("s1") is None


-class TestPruneSessions:
+class TestPruneWorkstreams:
    def test_orphan_removed(self, backend):
-        backend.register_session("orphan")
-        orphans, stale = backend.prune_sessions()
+        backend.register_workstream("orphan")
+        orphans, stale = backend.prune_workstreams()
        assert orphans == 1

    def test_stale_removed(self, backend):
        import sqlalchemy as sa

-        backend.register_session("old")
+        backend.register_workstream("old")
        backend.save_message("old", "user", "hi")
        # Force old timestamp
        with backend._engine.connect() as conn:
            conn.execute(
-                sa.text("UPDATE sessions SET updated = '2020-01-01' WHERE session_id = 'old'")
+                sa.text("UPDATE workstreams SET updated = '2020-01-01' WHERE ws_id = 'old'")
            )
            conn.commit()
-        _, stale = backend.prune_sessions(retention_days=30)
+        _, stale = backend.prune_workstreams(retention_days=30)
        assert stale == 1


-class TestResolveSession:
+class TestResolveWorkstream:
    def test_exact_alias(self, backend):
-        backend.register_session("s1")
-        backend.set_session_alias("s1", "myalias")
-        assert backend.resolve_session("myalias") == "s1"
+        backend.register_workstream("s1")
+        backend.set_workstream_alias("s1", "myalias")
+        assert backend.resolve_workstream("myalias") == "s1"

    def test_exact_id(self, backend):
-        backend.register_session("abc-123-def")
-        assert backend.resolve_session("abc-123-def") == "abc-123-def"
+        backend.register_workstream("abc-123-def")
+        assert backend.resolve_workstream("abc-123-def") == "abc-123-def"

    def test_prefix_match(self, backend):
-        backend.register_session("abc-123-def")
-        assert backend.resolve_session("abc") == "abc-123-def"
+        backend.register_workstream("abc-123-def")
+        assert backend.resolve_workstream("abc") == "abc-123-def"

    def test_not_found(self, backend):
-        assert backend.resolve_session("nonexistent") is None
+        assert backend.resolve_workstream("nonexistent") is None


-# -- Session config ------------------------------------------------------------
+# -- Workstream config ---------------------------------------------------------


-class TestSessionConfig:
+class TestWorkstreamConfig:
    def test_roundtrip(self, backend):
-        backend.register_session("s1")
-        backend.save_session_config("s1", {"temperature": "0.7", "effort": "high"})
-        cfg = backend.load_session_config("s1")
+        backend.register_workstream("s1")
+        backend.save_workstream_config("s1", {"temperature": "0.7", "effort": "high"})
+        cfg = backend.load_workstream_config("s1")
        assert cfg == {"temperature": "0.7", "effort": "high"}

    def test_empty_config(self, backend):
-        assert backend.load_session_config("nonexistent") == {}
+        assert backend.load_workstream_config("nonexistent") == {}


-# -- Session metadata ----------------------------------------------------------
+# -- Workstream metadata ------------------------------------------------------


-class TestSessionMetadata:
+class TestWorkstreamMetadata:
    def test_alias(self, backend):
-        backend.register_session("s1")
-        assert backend.set_session_alias("s1", "my-session")
-        assert backend.get_session_name("s1") == "my-session"
+        backend.register_workstream("s1")
+        assert backend.set_workstream_alias("s1", "my-session")
+        assert backend.get_workstream_display_name("s1") == "my-session"

    def test_alias_conflict(self, backend):
-        backend.register_session("s1")
-        backend.register_session("s2")
-        backend.set_session_alias("s1", "taken")
-        assert not backend.set_session_alias("s2", "taken")
+        backend.register_workstream("s1")
+        backend.register_workstream("s2")
+        backend.set_workstream_alias("s1", "taken")
+        assert not backend.set_workstream_alias("s2", "taken")

    def test_title(self, backend):
-        backend.register_session("s1")
-        backend.update_session_title("s1", "My Title")
-        assert backend.get_session_name("s1") == "My Title"
+        backend.register_workstream("s1")
+        backend.update_workstream_title("s1", "My Title")
+        assert backend.get_workstream_display_name("s1") == "My Title"

    def test_alias_preferred_over_title(self, backend):
-        backend.register_session("s1")
-        backend.update_session_title("s1", "Title")
-        backend.set_session_alias("s1", "Alias")
-        assert backend.get_session_name("s1") == "Alias"
+        backend.register_workstream("s1")
+        backend.update_workstream_title("s1", "Title")
+        backend.set_workstream_alias("s1", "Alias")
+        assert backend.get_workstream_display_name("s1") == "Alias"


 # -- Key-value store -----------------------------------------------------------
@@ -234,7 +234,7 @@ class TestKVStore:

 class TestSearch:
    def test_search_history(self, backend):
-        backend.register_session("s1")
+        backend.register_workstream("s1")
        backend.save_message("s1", "user", "hello world")
        backend.save_message("s1", "user", "goodbye world")
        results = backend.search_history("hello")
@@ -242,7 +242,7 @@ class TestSearch:
        assert any("hello" in str(r[3]) for r in results)

    def test_search_recent(self, backend):
-        backend.register_session("s1")
+        backend.register_workstream("s1")
        backend.save_message("s1", "user", "msg1")
        backend.save_message("s1", "user", "msg2")
        results = backend.search_history_recent(limit=1)
@@ -293,15 +293,14 @@ class TestWorkstreams:
        assert len(rows) == 1
        assert rows[0][0] == "ws1"

-    def test_session_with_ws_id(self, backend):
+    def test_workstream_with_messages_in_history(self, backend):
        backend.register_workstream("ws1", node_id="node-a")
-        backend.register_session("s1", node_id="node-a", ws_id="ws1")
-        backend.save_message("s1", "user", "hello")
-        rows = backend.list_sessions()
+        backend.save_message("ws1", "user", "hello")
+        rows = backend.list_workstreams_with_history()
        assert len(rows) == 1
-        # Columns: sid, alias, title, created, updated, count, node_id, ws_id
+        # Columns: ws_id, alias, title, created, updated, count, node_id
+        assert rows[0][0] == "ws1"
        assert rows[0][6] == "node-a"
-        assert rows[0][7] == "ws1"


 # -- Lifecycle -----------------------------------------------------------------
@@ -72,16 +72,16 @@ class TestToolsMetadata:
    """Validate the metadata extracted from JSON files."""

    def test_tool_count(self):
-        assert len(TOOLS) == 14
+        assert len(TOOLS) == 15

    def test_agent_tools_count(self):
-        assert len(AGENT_TOOLS) == 6
+        assert len(AGENT_TOOLS) == 7

    def test_task_agent_tools_count(self):
-        assert len(TASK_AGENT_TOOLS) == 9
+        assert len(TASK_AGENT_TOOLS) == 10

    def test_auto_approve_sets_match(self):
-        expected = {"read_file", "search", "math", "man", "web_fetch", "web_search"}
+        expected = {"read_file", "search", "math", "man", "web_fetch", "web_search", "notify"}
        assert expected == AGENT_AUTO_TOOLS
        assert expected == TASK_AUTO_TOOLS

@@ -101,6 +101,7 @@ class TestToolsMetadata:
            "remember": "key",
            "recall": "query",
            "forget": "key",
+            "notify": "message",
        }
        assert expected == PRIMARY_KEY_MAP

@@ -127,21 +127,7 @@ class TestApiTokenCRUD:
        assert "expires" not in tok


-class TestSessionWorkstreamUserId:
-    def test_register_session_with_user_id(self, db):
-        db.register_session("s1", user_id="u1")
-        # Verify via raw SQL that user_id is stored
-        import sqlalchemy as sa
-
-        from turnstone.core.storage._schema import sessions
-
-        with db._engine.connect() as conn:
-            row = conn.execute(
-                sa.select(sessions.c.user_id).where(sessions.c.session_id == "s1")
-            ).fetchone()
-            assert row is not None
-            assert row[0] == "u1"
-
+class TestWorkstreamUserId:
    def test_register_workstream_with_user_id(self, db):
        db.register_workstream("ws1", user_id="u1")
        import sqlalchemy as sa
@@ -155,15 +141,15 @@ class TestSessionWorkstreamUserId:
            assert row is not None
            assert row[0] == "u1"

-    def test_register_session_without_user_id(self, db):
-        db.register_session("s1")
+    def test_register_workstream_without_user_id(self, db):
+        db.register_workstream("ws1")
        import sqlalchemy as sa

-        from turnstone.core.storage._schema import sessions
+        from turnstone.core.storage._schema import workstreams

        with db._engine.connect() as conn:
            row = conn.execute(
-                sa.select(sessions.c.user_id).where(sessions.c.session_id == "s1")
+                sa.select(workstreams.c.user_id).where(workstreams.c.ws_id == "ws1")
            ).fetchone()
            assert row is not None
            assert row[0] is None
@@ -1,3 +1,3 @@
 """turnstone - Multi-node AI orchestration platform with tool use, agent routing, and cluster simulation."""

-__version__ = "0.4.0"
+__version__ = "0.4.3"
@@ -23,13 +23,18 @@ from turnstone.api.schemas import (
    AuthSetupRequest,
    AuthSetupResponse,
    AuthStatusResponse,
+    CreateScheduleRequest,
    CreateTokenRequest,
    CreateTokenResponse,
    CreateUserRequest,
    ErrorResponse,
+    ListScheduleRunsResponse,
+    ListSchedulesResponse,
    ListTokensResponse,
    ListUsersResponse,
+    ScheduleInfo,
    StatusResponse,
+    UpdateScheduleRequest,
    UserInfo,
 )

@@ -182,6 +187,61 @@ CONSOLE_ENDPOINTS: list[EndpointSpec] = [
        error_codes=[404],
        tags=["Admin"],
    ),
+    # --- Schedules ---
+    EndpointSpec(
+        "/v1/api/admin/schedules",
+        "GET",
+        "List all scheduled tasks",
+        response_model=ListSchedulesResponse,
+        tags=["Schedules"],
+    ),
+    EndpointSpec(
+        "/v1/api/admin/schedules",
+        "POST",
+        "Create a scheduled task",
+        request_model=CreateScheduleRequest,
+        response_model=ScheduleInfo,
+        error_codes=[400],
+        tags=["Schedules"],
+    ),
+    EndpointSpec(
+        "/v1/api/admin/schedules/{task_id}",
+        "GET",
+        "Get a scheduled task",
+        response_model=ScheduleInfo,
+        error_codes=[404],
+        tags=["Schedules"],
+    ),
+    EndpointSpec(
+        "/v1/api/admin/schedules/{task_id}",
+        "PUT",
+        "Update a scheduled task",
+        request_model=UpdateScheduleRequest,
+        response_model=ScheduleInfo,
+        error_codes=[400, 404],
+        tags=["Schedules"],
+    ),
+    EndpointSpec(
+        "/v1/api/admin/schedules/{task_id}",
+        "DELETE",
+        "Delete a scheduled task",
+        response_model=StatusResponse,
+        error_codes=[404],
+        tags=["Schedules"],
+    ),
+    EndpointSpec(
+        "/v1/api/admin/schedules/{task_id}/runs",
+        "GET",
+        "List run history for a scheduled task",
+        response_model=ListScheduleRunsResponse,
+        query_params=[
+            QueryParam(
+                "limit", "Max results (default 50, max 200)", schema_type="integer", default=50
+            ),
+        ],
+        error_codes=[404],
+        tags=["Schedules"],
+    ),
    # --- Observability ---
    EndpointSpec(
        "/health",
@@ -213,6 +273,11 @@ _ALL_MODELS: list[type[BaseModel]] = [
    ConsoleCreateWsRequest,
    ConsoleCreateWsResponse,
    ConsoleHealthResponse,
+    CreateScheduleRequest,
+    UpdateScheduleRequest,
+    ScheduleInfo,
+    ListSchedulesResponse,
+    ListScheduleRunsResponse,
 ]


@@ -155,3 +155,87 @@ class AuthStatusResponse(BaseModel):
    auth_enabled: bool
    has_users: bool
    setup_required: bool
+
+
+# ---------------------------------------------------------------------------
+# Schedules
+# ---------------------------------------------------------------------------
+
+
+class CreateScheduleRequest(BaseModel):
+    """POST /v1/api/admin/schedules request body."""
+
+    name: str = Field(description="Human-readable schedule name")
+    description: str = Field(default="", description="Optional description")
+    schedule_type: str = Field(description="'cron' or 'at'")
+    cron_expr: str = Field(default="", description="Cron expression (when schedule_type='cron')")
+    at_time: str = Field(default="", description="ISO8601 timestamp (when schedule_type='at')")
+    target_mode: str = Field(default="auto", description="auto, pool, all, or specific node_id")
+    model: str = Field(default="", description="Model alias for the workstream")
+    initial_message: str = Field(description="Message sent to the new workstream")
+    auto_approve: bool = Field(default=False)
+    auto_approve_tools: list[str] = Field(default_factory=list)
+    enabled: bool = Field(default=True)
+
+
+class UpdateScheduleRequest(BaseModel):
+    """PUT /v1/api/admin/schedules/{task_id} request body (partial update)."""
+
+    name: str | None = None
+    description: str | None = None
+    schedule_type: str | None = None
+    cron_expr: str | None = None
+    at_time: str | None = None
+    target_mode: str | None = None
+    model: str | None = None
+    initial_message: str | None = None
+    auto_approve: bool | None = None
+    auto_approve_tools: list[str] | None = None
+    enabled: bool | None = None
+
+
+class ScheduleInfo(BaseModel):
+    """Scheduled task details."""
+
+    task_id: str
+    name: str
+    description: str = ""
+    schedule_type: str
+    cron_expr: str = ""
+    at_time: str = ""
+    target_mode: str = "auto"
+    model: str = ""
+    initial_message: str
+    auto_approve: bool = False
+    auto_approve_tools: list[str] = Field(default_factory=list)
+    enabled: bool = True
+    created_by: str = ""
+    last_run: str | None = None
+    next_run: str | None = None
+    created: str = ""
+    updated: str = ""
+
+
+class ListSchedulesResponse(BaseModel):
+    """GET /v1/api/admin/schedules response."""
+
+    schedules: list[ScheduleInfo]
+
+
+class ScheduleRunInfo(BaseModel):
+    """Single execution record for a scheduled task."""
+
+    run_id: str
+    task_id: str
+    node_id: str = ""
+    ws_id: str = ""
+    correlation_id: str = ""
+    started: str
+    status: str = "dispatched"
+    error: str = ""
+
+
+class ListScheduleRunsResponse(BaseModel):
+    """GET /v1/api/admin/schedules/{task_id}/runs response."""
+
+    runs: list[ScheduleRunInfo]
@@ -39,18 +39,19 @@ class CreateWorkstreamRequest(BaseModel):
    name: str = Field(default="", description="Workstream display name (auto-generated if empty)")
    model: str = Field(default="", description="Model alias from registry")
    auto_approve: bool = Field(default=False, description="Auto-approve all tool calls")
-    resume_session: str = Field(
+    resume_ws: str = Field(
        default="",
-        description="Session ID to resume atomically during creation (empty = fresh start)",
+        description="Workstream ID to resume atomically during creation (empty = fresh start)",
    )


 class CreateWorkstreamResponse(BaseModel):
    ws_id: str = Field(description="Unique ID of the new workstream")
    name: str = Field(description="Assigned workstream name")
-    resumed: bool = Field(default=False, description="Whether a previous session was resumed")
-    session_id: str = Field(default="", description="Resolved session ID (set when resumed)")
-    message_count: int = Field(default=0, description="Number of messages in the resumed session")
+    resumed: bool = Field(default=False, description="Whether a previous workstream was resumed")
+    message_count: int = Field(
+        default=0, description="Number of messages in the resumed workstream"
+    )


 class CloseWorkstreamRequest(BaseModel):
@@ -66,7 +67,6 @@ class WorkstreamInfo(BaseModel):
    id: str
    name: str
    state: str
-    session_id: str | None = None


 class ListWorkstreamsResponse(BaseModel):
@@ -77,7 +77,6 @@ class DashboardWorkstream(BaseModel):
    id: str
    name: str
    state: str
-    session_id: str | None = None
    title: str = ""
    tokens: int = 0
    context_ratio: float = 0.0
@@ -104,12 +103,12 @@ class DashboardResponse(BaseModel):


 # ---------------------------------------------------------------------------
-# Sessions
+# Saved workstreams
 # ---------------------------------------------------------------------------


-class SessionInfo(BaseModel):
-    session_id: str
+class SavedWorkstreamInfo(BaseModel):
+    ws_id: str
    alias: str | None = None
    title: str | None = None
    created: str
@@ -117,8 +116,8 @@ class SessionInfo(BaseModel):
    message_count: int


-class ListSessionsResponse(BaseModel):
-    sessions: list[SessionInfo]
+class ListSavedWorkstreamsResponse(BaseModel):
+    workstreams: list[SavedWorkstreamInfo]


 # ---------------------------------------------------------------------------
@@ -25,7 +25,7 @@ from turnstone.api.server_schemas import (
    CreateWorkstreamResponse,
    DashboardResponse,
    HealthResponse,
-    ListSessionsResponse,
+    ListSavedWorkstreamsResponse,
    ListWorkstreamsResponse,
    PlanFeedbackRequest,
    SendRequest,
@@ -122,13 +122,13 @@ SERVER_ENDPOINTS: list[EndpointSpec] = [
        "across all workstreams. Returns text/event-stream.",
        tags=["Streaming"],
    ),
-    # --- Sessions ---
+    # --- Saved workstreams ---
    EndpointSpec(
-        "/v1/api/sessions",
+        "/v1/api/workstreams/saved",
        "GET",
-        "List saved sessions",
-        response_model=ListSessionsResponse,
-        tags=["Sessions"],
+        "List saved workstreams",
+        response_model=ListSavedWorkstreamsResponse,
+        tags=["Workstreams"],
    ),
    # --- Auth ---
    EndpointSpec(
@@ -191,7 +191,7 @@ _ALL_MODELS: list[type[BaseModel]] = [
    CloseWorkstreamRequest,
    ListWorkstreamsResponse,
    DashboardResponse,
-    ListSessionsResponse,
+    ListSavedWorkstreamsResponse,
    HealthResponse,
 ]

@@ -0,0 +1,195 @@
+"""Lightweight HTTP server for the channel gateway.
+
+Runs alongside the channel adapters (Discord, etc.) to receive notification
+requests from the bridge.  Exposes ``POST /v1/api/notify`` and ``GET /health``.
+"""
+
+from __future__ import annotations
+
+import asyncio
+import json
+import socket
+import uuid
+from typing import TYPE_CHECKING, Any
+
+from starlette.applications import Starlette
+from starlette.responses import JSONResponse
+from starlette.routing import Mount, Route
+
+from turnstone.core.log import get_logger
+
+if TYPE_CHECKING:
+    from starlette.requests import Request
+
+    from turnstone.channels._protocol import ChannelAdapter
+    from turnstone.core.storage._protocol import StorageBackend
+
+log = get_logger(__name__)
+
+
+async def _handle_health(request: Request) -> JSONResponse:
+    return JSONResponse({"status": "ok", "service": "channel"})
+
+
+def _check_auth(request: Request) -> JSONResponse | None:
+    """Validate the request's Authorization header.  Returns an error response or None."""
+    auth_token: str = getattr(request.app.state, "auth_token", "")
+    jwt_secret: str = getattr(request.app.state, "jwt_secret", "")
+
+    if not auth_token and not jwt_secret:
+        log.warning("notify.auth_not_configured")
+        return JSONResponse({"error": "authentication not configured"}, status_code=401)
+
+    header = request.headers.get("Authorization", "")
+    if not header.startswith("Bearer "):
+        return JSONResponse({"error": "Unauthorized"}, status_code=401)
+
+    token = header[7:]
+
+    # Static token check
+    if auth_token:
+        import hmac
+
+        if hmac.compare_digest(token, auth_token):
+            return None
+
+    # JWT check
+    if jwt_secret and "." in token:
+        from turnstone.core.auth import JWT_AUD_CHANNEL, validate_jwt
+
+        result = validate_jwt(token, jwt_secret, audience=JWT_AUD_CHANNEL)
+        if result is not None:
+            return None
+
+    return JSONResponse({"error": "Unauthorized"}, status_code=401)
+
+
+async def _handle_notify(request: Request) -> JSONResponse:
+    """Deliver a notification to one or more channel adapters."""
+    auth_err = _check_auth(request)
+    if auth_err is not None:
+        return auth_err
+
+    adapters: dict[str, ChannelAdapter] = request.app.state.adapters
+    storage: StorageBackend = request.app.state.storage
+
+    try:
+        body: dict[str, Any] = await request.json()
+    except (json.JSONDecodeError, ValueError):
+        return JSONResponse({"error": "invalid JSON"}, status_code=400)
+
+    target = body.get("target")
+    message = body.get("message", "").strip() if isinstance(body.get("message"), str) else ""
+    title = body.get("title", "").strip() if isinstance(body.get("title"), str) else ""
+
+    if not target or not message:
+        return JSONResponse({"error": "target and message are required"}, status_code=400)
+
+    content = f"**{title}**\n{message}" if title else message
+
+    # Resolve targets
+    targets: list[tuple[str, str]] = []
+    if "username" in target:
+        user = await asyncio.to_thread(storage.get_user_by_username, target["username"])
+        if user is None:
+            log.warning("notify.user_not_found", username=target["username"])
+            return JSONResponse(
+                {"error": "target not found or has no linked channels"},
+                status_code=404,
+            )
+        links = await asyncio.to_thread(storage.list_channel_users_by_user, user["user_id"])
+        for link in links:
+            targets.append((link["channel_type"], link["channel_user_id"]))
+        if not targets:
+            log.warning("notify.user_no_linked_channels", username=target["username"])
+            return JSONResponse(
+                {"error": "target not found or has no linked channels"},
+                status_code=404,
+            )
+    elif "channel_type" in target and "channel_id" in target:
+        targets.append((target["channel_type"], target["channel_id"]))
+    else:
+        return JSONResponse(
+            {"error": "target must have username or channel_type+channel_id"},
+            status_code=400,
+        )
+
+    results: list[dict[str, str]] = []
+    for channel_type, channel_id in targets:
+        adapter = adapters.get(channel_type)
+        if adapter is None:
+            results.append(
+                {
+                    "channel_type": channel_type,
+                    "channel_id": channel_id,
+                    "status": "no_adapter",
+                }
+            )
+            log.warning(
+                "notify.no_adapter",
+                channel_type=channel_type,
+                channel_id=channel_id,
+            )
+            continue
+        try:
+            msg_id = await adapter.send(channel_id, content)
+            results.append(
+                {
+                    "channel_type": channel_type,
+                    "channel_id": channel_id,
+                    "status": "sent",
+                    "message_id": msg_id,
+                }
+            )
+            log.info(
+                "notify.delivered",
+                channel_type=channel_type,
+                channel_id=channel_id,
+                message_id=msg_id,
+            )
+        except Exception:
+            log.exception(
+                "notify.delivery_failed",
+                channel_type=channel_type,
+                channel_id=channel_id,
+            )
+            results.append(
+                {
+                    "channel_type": channel_type,
+                    "channel_id": channel_id,
+                    "status": "failed",
+                }
+            )
+
+    return JSONResponse({"results": results})
+
+
+def create_channel_app(
+    adapters: dict[str, ChannelAdapter],
+    storage: StorageBackend,
+    *,
+    auth_token: str = "",
+    jwt_secret: str = "",
+) -> Starlette:
+    """Create the channel gateway HTTP application."""
+    app = Starlette(
+        routes=[
+            Route("/health", _handle_health),
+            Mount(
+                "/v1",
+                routes=[
+                    Route("/api/notify", _handle_notify, methods=["POST"]),
+                ],
+            ),
+        ],
+    )
+    app.state.adapters = adapters
+    app.state.storage = storage
+    app.state.auth_token = auth_token
+    app.state.jwt_secret = jwt_secret
+    return app
+
+
+def _get_service_id() -> str:
+    """Generate a unique service ID from hostname + random suffix."""
+    return f"channel-{socket.gethostname()}-{uuid.uuid4().hex[:8]}"
@@ -150,7 +150,7 @@ class ChannelRouter:
                owner = await self._broker.get_ws_owner(route["ws_id"])
                if owner:
                    return route["ws_id"], False
-                # Workstream was evicted/closed — capture old ws_id for session
+                # Workstream was evicted/closed — capture old ws_id for
                # resume, then remove the stale route.
                old_ws_id = route["ws_id"]
                await asyncio.to_thread(
@@ -163,20 +163,13 @@ class ChannelRouter:
                    channel_id=channel_id,
                )

-            # 2. Look up old session for atomic resume (if stale route).
-            resume_session = ""
-            if old_ws_id:
-                old_sid: str | None = await asyncio.to_thread(
-                    self._storage.get_session_id_by_ws, old_ws_id
-                )
-                resume_session = old_sid or ""
-
-            # 3. Create via MQ with atomic resume.
+            # 2. Create via MQ with atomic resume (reuse old ws_id directly).
+            resume_ws = old_ws_id or ""
            msg = CreateWorkstreamMessage(
                name=name,
                model=model,
-                initial_message="" if resume_session else initial_message,
-                resume_session=resume_session,
+                initial_message="" if resume_ws else initial_message,
+                resume_ws=resume_ws,
                auto_approve=self._auto_approve,
                auto_approve_tools=list(self._auto_approve_tools),
            )
@@ -190,7 +183,7 @@ class ChannelRouter:
                correlation_id=cid,
                channel_type=channel_type,
                channel_id=channel_id,
-                resume_session=resume_session or None,
+                resume_ws=resume_ws or None,
            )

            try:
@@ -1,8 +1,8 @@
 """Unified channel gateway entry point.

 Launches one or more channel adapters (Discord, Slack, etc.) connected to
-the turnstone cluster via Redis MQ.  Currently supports Discord; future
-adapters will be added as additional ``--*-token`` flags.
+the turnstone cluster via Redis MQ.  An HTTP server runs alongside for
+inbound notification delivery from the server.

 Run as: ``turnstone-channel --discord-token $TURNSTONE_DISCORD_TOKEN``
 """
@@ -10,6 +10,7 @@ Run as: ``turnstone-channel --discord-token $TURNSTONE_DISCORD_TOKEN``
 from __future__ import annotations

 import os
+import socket
 import sys


@@ -44,6 +45,26 @@ def main() -> None:
        help="Comma-separated list of allowed Discord channel IDs (default: all)",
    )

+    # -- HTTP server ---------------------------------------------------------
+    parser.add_argument(
+        "--http-host",
+        default="127.0.0.1",
+        help="HTTP server bind address (default: %(default)s)",
+    )
+    parser.add_argument(
+        "--http-port",
+        type=int,
+        default=int(os.environ.get("TURNSTONE_CHANNEL_PORT", "8091")),
+        help="HTTP server port (default: $TURNSTONE_CHANNEL_PORT or 8091)",
+    )
+
+    # -- Auth ----------------------------------------------------------------
+    parser.add_argument(
+        "--auth-token",
+        default=os.environ.get("TURNSTONE_CHANNEL_AUTH_TOKEN", ""),
+        help="Static auth token for /v1/api/notify (default: $TURNSTONE_CHANNEL_AUTH_TOKEN)",
+    )
+
    # -- Workstream defaults -------------------------------------------------
    parser.add_argument(
        "--model",
@@ -85,6 +106,10 @@ def main() -> None:
        path=db_path,
    )

+    # -- Auth config ---------------------------------------------------------
+    auth_token = args.auth_token
+    jwt_secret = os.environ.get("TURNSTONE_JWT_SECRET", "").strip()
+
    # -- Broker --------------------------------------------------------------
    from turnstone.mq.broker import async_broker_from_args

@@ -106,6 +131,9 @@ def main() -> None:

    # -- Run -----------------------------------------------------------------
    if args.discord_token:
+        import asyncio
+
+        from turnstone.channels._http import _get_service_id, create_channel_app
        from turnstone.channels.discord.bot import TurnstoneBot
        from turnstone.channels.discord.config import DiscordConfig
        from turnstone.core.storage._registry import get_storage
@@ -128,9 +156,82 @@ def main() -> None:
            allowed_channels=allowed_channels,
        )

-        bot = TurnstoneBot(config, broker, get_storage())
-        log.info("channel.starting", adapter="discord", guild_id=config.guild_id)
-        bot.run()
+        storage = get_storage()
+        bot = TurnstoneBot(config, broker, storage)
+        adapters = {"discord": bot}
+
+        # Create HTTP app for notification delivery
+        channel_app = create_channel_app(
+            adapters,  # type: ignore[arg-type]
+            storage,
+            auth_token=auth_token,
+            jwt_secret=jwt_secret,
+        )
+
+        log.info(
+            "channel.starting",
+            adapter="discord",
+            guild_id=config.guild_id,
+            http_port=args.http_port,
+        )
+
+        async def _run_all() -> None:
+            """Run Discord bot + HTTP server + service heartbeat concurrently."""
+            import uvicorn
+
+            service_id = _get_service_id()
+
+            # Resolve advertise URL — env override for Docker/K8s,
+            # otherwise derive from bind address.
+            advertise_url = os.environ.get("TURNSTONE_CHANNEL_ADVERTISE_URL", "").strip()
+            if not advertise_url:
+                if args.http_host in ("0.0.0.0", "::"):
+                    advertise_host = socket.gethostname()
+                else:
+                    advertise_host = args.http_host
+                advertise_url = f"http://{advertise_host}:{args.http_port}"
+            service_url = advertise_url
+
+            # Register in service registry
+            storage.register_service("channel", service_id, service_url)
+            log.info(
+                "channel.service_registered",
+                service_id=service_id,
+                url=service_url,
+            )
+
+            async def _heartbeat_loop() -> None:
+                """Periodically update service heartbeat."""
+                while True:
+                    await asyncio.sleep(30)
+                    try:
+                        await asyncio.to_thread(storage.heartbeat_service, "channel", service_id)
+                    except Exception:
+                        log.exception("channel.heartbeat_failed")
+
+            uv_config = uvicorn.Config(
+                channel_app,
+                host=args.http_host,
+                port=args.http_port,
+                log_level="warning",
+            )
+            server = uvicorn.Server(uv_config)
+
+            heartbeat_task = asyncio.create_task(_heartbeat_loop())
+            try:
+                await asyncio.gather(
+                    bot.start(),
+                    server.serve(),
+                )
+            finally:
+                heartbeat_task.cancel()
+                await asyncio.to_thread(storage.deregister_service, "channel", service_id)
+                log.info("channel.service_deregistered", service_id=service_id)
+
+        import contextlib
+
+        with contextlib.suppress(KeyboardInterrupt):
+            asyncio.run(_run_all())


 if __name__ == "__main__":
@@ -21,8 +21,8 @@ from turnstone.mq.protocol import (
    ErrorEvent,
    OutboundEvent,
    PlanReviewEvent,
-    SessionResumedEvent,
    TurnCompleteEvent,
+    WorkstreamResumedEvent,
 )

 if TYPE_CHECKING:
@@ -307,10 +307,10 @@ class TurnstoneBot:
            if sm is not None:
                await sm.finalize()

-        elif isinstance(event, SessionResumedEvent):
-            name = event.name or "previous session"
+        elif isinstance(event, WorkstreamResumedEvent):
+            name = event.name or "previous workstream"
            count = event.message_count
-            await thread.send(f"*Session resumed: {name} ({count} messages restored)*")
+            await thread.send(f"*Resumed: {name} ({count} messages restored)*")

        elif isinstance(event, ErrorEvent):
            safe_msg = event.message[:500] if event.message else "An error occurred"
@@ -348,6 +348,32 @@ class TurnstoneBot:
        """Start the bot (async). Use this for multi-adapter ``asyncio.gather``."""
        await self._bot.start(self.config.bot_token, reconnect=True)

+    async def send(self, channel_id: str, content: str) -> str:
+        """Send a message to a Discord channel or user DM.
+
+        Implements the :class:`ChannelAdapter` protocol.  Tries the ID as a
+        channel first; if not found, attempts a user DM.  Long messages are
+        chunked via :func:`chunk_message`.
+        """
+        import discord
+
+        int_id = int(channel_id)
+        target: discord.abc.Messageable | None = self._bot.get_channel(int_id)  # type: ignore[assignment]
+        if target is None:
+            try:
+                user = await self._bot.fetch_user(int_id)
+                target = await user.create_dm()
+            except discord.NotFound as exc:
+                raise ValueError(f"Discord channel/user {channel_id} not found") from exc
+
+        content = discord.utils.escape_mentions(content)
+        chunks = chunk_message(content, self.config.max_message_length)
+        msg: discord.Message | None = None
+        for chunk in chunks:
+            msg = await target.send(chunk)  # type: ignore[union-attr]
+
+        return str(msg.id) if msg else ""
+
    async def stop(self) -> None:
        """Disconnect the bot and clean up subscriptions."""
        for ws_id in list(self._subscribed_ws):
@@ -41,7 +41,7 @@ SLASH_COMMANDS = [
    "/instructions",
    "/clear",
    "/new",
-    "/sessions",
+    "/workstreams",
    "/resume",
    "/name",
    "/delete",
@@ -787,8 +787,8 @@ def main() -> None:
    parser.add_argument(
        "--resume",
        default=None,
-        metavar="SESSION",
-        help="Resume a previous session by alias or session_id",
+        metavar="WS",
+        help="Resume a previous workstream by alias or ws_id",
    )
    parser.add_argument(
        "--skip-permissions",
@@ -801,11 +801,11 @@ def main() -> None:
        help="API key (default: $OPENAI_API_KEY, or 'dummy' for local servers)",
    )
    parser.add_argument(
-        "--session-retention-days",
+        "--retention-days",
        type=int,
        default=90,
        metavar="DAYS",
-        help="Delete unnamed sessions older than DAYS days on startup, 0 to disable (default: 90)",
+        help="Delete unnamed workstreams older than DAYS days on startup, 0 to disable (default: 90)",
    )
    parser.add_argument(
        "--console-url",
@@ -845,10 +845,10 @@ def main() -> None:
    )
    init_storage(db_backend, path=db_path, url=db_url, pool_size=db_pool_size)

-    # Prune stale / empty sessions on startup
-    from turnstone.core.memory import prune_sessions
+    # Prune stale / empty workstreams on startup
+    from turnstone.core.memory import prune_workstreams

-    prune_sessions(retention_days=args.session_retention_days, log_fn=print)
+    prune_workstreams(retention_days=args.retention_days, log_fn=print)

    # Set up readline
    setup_readline()
@@ -894,7 +894,7 @@ def main() -> None:

    mcp_client = create_mcp_client(getattr(args, "mcp_config", None))

-    # Session factory — captures shared config for creating workstream sessions
+    # ChatSession factory — captures shared config for creating workstreams
    def session_factory(
        ui: SessionUI | None, model_alias: str | None = None, ws_id: str | None = None
    ) -> ChatSession:
@@ -929,19 +929,19 @@ def main() -> None:

    # Handle --resume
    if args.resume:
-        from turnstone.core.memory import resolve_session
+        from turnstone.core.memory import resolve_workstream

-        target_id = resolve_session(args.resume)
+        target_id = resolve_workstream(args.resume)
        if not target_id:
-            print(red(f"Session not found: {args.resume}"))
+            print(red(f"Workstream not found: {args.resume}"))
            sys.exit(1)
        if ws.session is None:
            print(red("No session available."))
            sys.exit(1)
-        if not ws.session.resume_session(target_id):
-            print(red(f"Session '{args.resume}' has no messages."))
+        if not ws.session.resume(target_id):
+            print(red(f"Workstream '{args.resume}' has no messages."))
            sys.exit(1)
-        print(f"Resumed session {bold(target_id)} ({len(ws.session.messages)} messages)")
+        print(f"Resumed workstream {bold(target_id)} ({len(ws.session.messages)} messages)")

    # Background attention notification — write to stderr while user types
    def _bg_attention_notify(ws_id: str, state: WorkstreamState) -> None:
@@ -0,0 +1,259 @@
+"""Background task scheduler for timed workstream dispatch.
+
+Runs as a daemon thread inside the console process. Checks for due tasks
+every ``check_interval`` seconds and dispatches them as
+``CreateWorkstreamMessage`` via the MQ broker.
+
+Uses Redis ``SET NX EX`` for distributed locking in multi-console deployments.
+"""
+
+from __future__ import annotations
+
+import threading
+import uuid
+from datetime import UTC, datetime
+from typing import TYPE_CHECKING, Any
+
+import structlog
+
+if TYPE_CHECKING:
+    from turnstone.console.collector import ClusterCollector
+    from turnstone.core.storage._protocol import StorageBackend
+    from turnstone.mq.broker import RedisBroker
+
+log = structlog.get_logger(__name__)
+
+
+def _pick_best_node(collector: ClusterCollector) -> str:
+    """Select the reachable node with the most available capacity."""
+    nodes, _ = collector.get_nodes(sort_by="activity", limit=1000, offset=0)
+    best_id = ""
+    best_headroom = -1
+    for n in nodes:
+        if not n.get("reachable", False):
+            continue
+        headroom = n.get("max_ws", 10) - n.get("ws_total", 0)
+        if headroom > best_headroom:
+            best_headroom = headroom
+            best_id = n["node_id"]
+    return best_id
+
+
+class TaskScheduler:
+    """Background scheduler for dispatching timed workstreams."""
+
+    def __init__(
+        self,
+        broker: RedisBroker,
+        collector: ClusterCollector,
+        storage: StorageBackend,
+        prefix: str = "turnstone",
+        check_interval: float = 15.0,
+        lock_ttl: int = 60,
+        max_fan_out: int = 20,
+    ) -> None:
+        self._broker = broker
+        self._collector = collector
+        self._storage = storage
+        self._prefix = prefix
+        self._check_interval = check_interval
+        self._lock_ttl = lock_ttl
+        self._max_fan_out = max_fan_out
+        self._stop_event = threading.Event()
+        self._thread: threading.Thread | None = None
+        self._tick_count = 0
+        self._prune_every = 240  # ~1 hour at 15s intervals
+
+    def start(self) -> None:
+        """Start the scheduler daemon thread."""
+        self._stop_event.clear()
+        self._thread = threading.Thread(target=self._loop, daemon=True, name="scheduler")
+        self._thread.start()
+        log.info("scheduler.started", check_interval=self._check_interval)
+
+    def stop(self) -> None:
+        """Stop the scheduler and wait for the thread to finish."""
+        self._stop_event.set()
+        if self._thread is not None:
+            self._thread.join(timeout=5)
+        log.info("scheduler.stopped")
+
+    def _loop(self) -> None:
+        """Main scheduler loop — tick then sleep."""
+        while not self._stop_event.is_set():
+            try:
+                self._tick()
+            except Exception:
+                log.exception("scheduler.tick_error")
+            self._stop_event.wait(self._check_interval)
+
+    # Lua script for safe lock release — only delete if we still own the lock
+    _UNLOCK_SCRIPT = "if redis.call('get',KEYS[1])==ARGV[1] then return redis.call('del',KEYS[1]) else return 0 end"
+
+    def _tick(self) -> None:
+        """Single scheduler iteration: acquire lock, query due tasks, dispatch."""
+        now = datetime.now(UTC).strftime("%Y-%m-%dT%H:%M:%S")
+
+        # Distributed lock with unique owner — prevents releasing another instance's lock
+        lock_key = f"{self._prefix}:scheduler:lock"
+        lock_value = uuid.uuid4().hex
+        acquired = self._broker._redis.set(lock_key, lock_value, nx=True, ex=self._lock_ttl)
+        if not acquired:
+            return
+
+        try:
+            due_tasks = self._storage.list_due_tasks(now)
+            for task in due_tasks:
+                self._dispatch_task(task, now)
+
+            # Periodic run history pruning (~once per hour)
+            self._tick_count += 1
+            if self._tick_count % self._prune_every == 0:
+                pruned = self._storage.prune_task_runs(retention_days=90)
+                if pruned:
+                    log.info("scheduler.pruned_runs", count=pruned)
+        finally:
+            # Only release our own lock (safe even if TTL expired and another took it)
+            self._broker._redis.eval(  # type: ignore[no-untyped-call]
+                self._UNLOCK_SCRIPT, 1, lock_key, lock_value
+            )
+
+    def _dispatch_task(self, task: dict[str, Any], now: str) -> None:
+        """Dispatch a single task as one or more CreateWorkstreamMessages."""
+        target_mode = task["target_mode"]
+        task_id = task["task_id"]
+        dispatched = False
+
+        if target_mode == "all":
+            nodes, _ = self._collector.get_nodes(sort_by="activity", limit=1000, offset=0)
+            fan_count = 0
+            for n in nodes:
+                if n.get("reachable", False):
+                    if fan_count >= self._max_fan_out:
+                        log.warning(
+                            "scheduler.fan_out_capped",
+                            task_id=task_id,
+                            max_fan_out=self._max_fan_out,
+                        )
+                        break
+                    self._dispatch_to_node(task, n["node_id"], now)
+                    fan_count += 1
+                    dispatched = True
+            if not dispatched:
+                self._record_failure(task, now, "No reachable nodes for fan-out")
+        elif target_mode == "pool":
+            self._dispatch_to_pool(task, now)
+            dispatched = True
+        elif target_mode == "auto":
+            node_id = _pick_best_node(self._collector)
+            if node_id:
+                self._dispatch_to_node(task, node_id, now)
+                dispatched = True
+            else:
+                self._record_failure(task, now, "No reachable nodes")
+        else:
+            # Specific node_id
+            self._dispatch_to_node(task, target_mode, now)
+            dispatched = True
+
+        if not dispatched:
+            return  # Don't advance schedule on failure
+
+        # Update last_run and compute next_run
+        next_run = self._compute_next_run(task)
+        if task["schedule_type"] == "at":
+            self._storage.update_scheduled_task(task_id, last_run=now, next_run="", enabled=False)
+        else:
+            self._storage.update_scheduled_task(task_id, last_run=now, next_run=next_run)
+
+        log_kw: dict[str, Any] = {
+            "task_id": task_id,
+            "target_mode": target_mode,
+            "schedule_type": task["schedule_type"],
+            "created_by": task.get("created_by", ""),
+        }
+        if task.get("auto_approve", 0):
+            log_kw["auto_approve"] = True
+            log_kw["auto_approve_tools"] = task.get("auto_approve_tools", "")
+            log.warning("scheduler.task_dispatched_auto_approve", **log_kw)
+        else:
+            log.info("scheduler.task_dispatched", **log_kw)
+
+    @staticmethod
+    def _parse_tools(task: dict[str, Any]) -> list[str]:
+        raw = task.get("auto_approve_tools", "")
+        return [t.strip() for t in raw.split(",") if t.strip()]
+
+    def _dispatch_to_node(self, task: dict[str, Any], node_id: str, now: str) -> None:
+        """Send a CreateWorkstreamMessage to a specific node."""
+        from turnstone.mq.protocol import CreateWorkstreamMessage
+
+        msg = CreateWorkstreamMessage(
+            name=task["name"],
+            model=task.get("model", ""),
+            target_node=node_id,
+            initial_message=task["initial_message"],
+            auto_approve=bool(task.get("auto_approve", 0)),
+            auto_approve_tools=self._parse_tools(task),
+            user_id=task.get("created_by", ""),
+        )
+        self._broker.push_inbound(msg.to_json(), node_id=node_id)
+
+        self._storage.record_task_run(
+            run_id=uuid.uuid4().hex,
+            task_id=task["task_id"],
+            node_id=node_id,
+            ws_id="",
+            correlation_id=msg.correlation_id,
+            started=now,
+            status="dispatched",
+            error="",
+        )
+
+    def _dispatch_to_pool(self, task: dict[str, Any], now: str) -> None:
+        """Send a CreateWorkstreamMessage to the shared pool queue."""
+        from turnstone.mq.protocol import CreateWorkstreamMessage
+
+        msg = CreateWorkstreamMessage(
+            name=task["name"],
+            model=task.get("model", ""),
+            initial_message=task["initial_message"],
+            auto_approve=bool(task.get("auto_approve", 0)),
+            auto_approve_tools=self._parse_tools(task),
+            user_id=task.get("created_by", ""),
+        )
+        self._broker.push_inbound(msg.to_json())
+
+        self._storage.record_task_run(
+            run_id=uuid.uuid4().hex,
+            task_id=task["task_id"],
+            node_id="pool",
+            ws_id="",
+            correlation_id=msg.correlation_id,
+            started=now,
+            status="dispatched",
+            error="",
+        )
+
+    def _record_failure(self, task: dict[str, Any], now: str, error: str) -> None:
+        """Record a failed dispatch attempt."""
+        self._storage.record_task_run(
+            run_id=uuid.uuid4().hex,
+            task_id=task["task_id"],
+            node_id="",
+            ws_id="",
+            correlation_id="",
+            started=now,
+            status="failed",
+            error=error,
+        )
+        log.warning("scheduler.dispatch_failed", task_id=task["task_id"], error=error)
+
+    @staticmethod
+    def _compute_next_run(task: dict[str, Any]) -> str:
+        """Compute the next run time. Returns empty string for one-shot tasks."""
+        from turnstone.console.server import _compute_next_run
+
+        return _compute_next_run(
+            task["schedule_type"], task.get("cron_expr", ""), task.get("at_time", "")
+        )
@@ -620,8 +620,14 @@ async def _lifespan(app: Starlette) -> AsyncGenerator[None, None]:
        timeout=httpx.Timeout(connect=5, read=30, write=5, pool=5),
        headers=headers,
    )
+    # Start scheduler if configured
+    scheduler = getattr(app.state, "scheduler", None)
+    if scheduler is not None:
+        scheduler.start()
    yield
    # Shutdown
+    if scheduler is not None:
+        scheduler.stop()
    await app.state.proxy_sse_client.aclose()
    await app.state.proxy_client.aclose()
    app.state.collector.stop()
@@ -878,6 +884,277 @@ async def admin_delete_channel(request: Request) -> JSONResponse:
    return JSONResponse({"error": "Channel link not found"}, status_code=404)


+# ---------------------------------------------------------------------------
+# Admin API endpoints — scheduled tasks
+# ---------------------------------------------------------------------------
+
+
+def _normalize_task_dict(task: dict[str, Any]) -> dict[str, Any]:
+    """Convert DB row ints/csv to JSON-friendly bools/lists."""
+    tools_str = task.get("auto_approve_tools", "")
+    task["auto_approve_tools"] = [s.strip() for s in tools_str.split(",") if s.strip()]
+    task["auto_approve"] = bool(task.get("auto_approve", 0))
+    task["enabled"] = bool(task.get("enabled", 1))
+    return task
+
+
+def _compute_next_run(schedule_type: str, cron_expr: str, at_time: str) -> str:
+    """Compute the next run time for a schedule. Empty string if invalid."""
+    if schedule_type == "at":
+        return at_time
+    if schedule_type == "cron" and cron_expr:
+        from datetime import UTC, datetime
+
+        from croniter import croniter
+
+        cron = croniter(cron_expr, datetime.now(UTC))
+        next_dt = cron.get_next(datetime)
+        return str(next_dt.strftime("%Y-%m-%dT%H:%M:%S"))
+    return ""
+
+
+def _validate_schedule_fields(schedule_type: str, cron_expr: str, at_time: str) -> str | None:
+    """Validate schedule type/expression. Returns error string or None."""
+    if schedule_type not in ("cron", "at"):
+        return "schedule_type must be 'cron' or 'at'"
+    if schedule_type == "cron":
+        if not cron_expr:
+            return "cron_expr is required when schedule_type is 'cron'"
+        from croniter import croniter
+
+        if not croniter.is_valid(cron_expr):
+            return f"Invalid cron expression: {cron_expr}"
+    if schedule_type == "at":
+        if not at_time:
+            return "at_time is required when schedule_type is 'at'"
+        from datetime import UTC, datetime
+
+        try:
+            dt = datetime.fromisoformat(at_time)
+            if dt.tzinfo is None:
+                return (
+                    "at_time must include a timezone offset (e.g. 2024-01-01T12:00:00Z or +00:00)"
+                )
+            if dt <= datetime.now(UTC):
+                return "at_time must be in the future"
+        except ValueError:
+            return "at_time must be a valid ISO8601 timestamp with timezone"
+    return None
+
+
+async def admin_list_schedules(request: Request) -> JSONResponse:
+    """GET /v1/api/admin/schedules — list all scheduled tasks."""
+    from turnstone.core.web_helpers import require_storage_or_503
+
+    storage, err = require_storage_or_503(request)
+    if err:
+        return err
+    tasks = storage.list_scheduled_tasks()
+    for t in tasks:
+        _normalize_task_dict(t)
+    return JSONResponse({"schedules": tasks})
+
+
+async def admin_create_schedule(request: Request) -> JSONResponse:
+    """POST /v1/api/admin/schedules — create a scheduled task."""
+    import uuid
+
+    from turnstone.core.web_helpers import read_json_or_400, require_storage_or_503
+
+    storage, err = require_storage_or_503(request)
+    if err:
+        return err
+
+    body = await read_json_or_400(request)
+    if isinstance(body, JSONResponse):
+        return body
+
+    name = str(body.get("name", "")).strip()[:256]
+    description = str(body.get("description", "")).strip()[:1024]
+    schedule_type = str(body.get("schedule_type", "")).strip()
+    cron_expr = str(body.get("cron_expr", "")).strip()[:256]
+    at_time = str(body.get("at_time", "")).strip()[:64]
+    target_mode = str(body.get("target_mode", "auto")).strip()[:256]
+    model = str(body.get("model", "")).strip()[:128]
+    initial_message = str(body.get("initial_message", "")).strip()[:4096]
+    auto_approve = bool(body.get("auto_approve", False))
+    raw_tools = body.get("auto_approve_tools", [])
+    auto_approve_tools = raw_tools if isinstance(raw_tools, list) else []
+    enabled = bool(body.get("enabled", True))
+
+    if not name:
+        return JSONResponse({"error": "name is required"}, status_code=400)
+    if not initial_message:
+        return JSONResponse({"error": "initial_message is required"}, status_code=400)
+
+    validation_err = _validate_schedule_fields(schedule_type, cron_expr, at_time)
+    if validation_err:
+        return JSONResponse({"error": validation_err}, status_code=400)
+
+    if not target_mode:
+        return JSONResponse({"error": "target_mode is required"}, status_code=400)
+
+    # Cap total schedule count to prevent unbounded growth
+    max_schedules = 200
+    existing = storage.list_scheduled_tasks()
+    if len(existing) >= max_schedules:
+        return JSONResponse(
+            {"error": f"Maximum of {max_schedules} schedules reached"}, status_code=409
+        )
+
+    next_run = _compute_next_run(schedule_type, cron_expr, at_time)
+    task_id = uuid.uuid4().hex
+    created_by = getattr(getattr(request, "state", None), "user_id", "")
+
+    storage.create_scheduled_task(
+        task_id=task_id,
+        name=name,
+        description=description,
+        schedule_type=schedule_type,
+        cron_expr=cron_expr,
+        at_time=at_time,
+        target_mode=target_mode,
+        model=model,
+        initial_message=initial_message,
+        auto_approve=auto_approve,
+        auto_approve_tools=auto_approve_tools,
+        created_by=created_by,
+        next_run=next_run if enabled else "",
+    )
+
+    if not enabled:
+        # Storage backends default enabled=1 on create; persist user's choice
+        storage.update_scheduled_task(task_id, enabled=False)
+
+    task = storage.get_scheduled_task(task_id)
+    if task:
+        _normalize_task_dict(task)
+    return JSONResponse(task)
+
+
+async def admin_get_schedule(request: Request) -> JSONResponse:
+    """GET /v1/api/admin/schedules/{task_id} — get single task."""
+    from turnstone.core.web_helpers import require_storage_or_503
+
+    storage, err = require_storage_or_503(request)
+    if err:
+        return err
+    task_id = request.path_params["task_id"]
+    task = storage.get_scheduled_task(task_id)
+    if task is None:
+        return JSONResponse({"error": "Schedule not found"}, status_code=404)
+    _normalize_task_dict(task)
+    return JSONResponse(task)
+
+
+async def admin_update_schedule(request: Request) -> JSONResponse:
+    """PUT /v1/api/admin/schedules/{task_id} — partial update."""
+    from turnstone.core.web_helpers import read_json_or_400, require_storage_or_503
+
+    storage, err = require_storage_or_503(request)
+    if err:
+        return err
+    task_id = request.path_params["task_id"]
+
+    existing = storage.get_scheduled_task(task_id)
+    if existing is None:
+        return JSONResponse({"error": "Schedule not found"}, status_code=404)
+
+    body = await read_json_or_400(request)
+    if isinstance(body, JSONResponse):
+        return body
+
+    updates: dict[str, Any] = {}
+    if "name" in body:
+        updates["name"] = str(body["name"]).strip()[:256]
+    if "description" in body:
+        updates["description"] = str(body["description"]).strip()[:1024]
+    if "schedule_type" in body:
+        updates["schedule_type"] = str(body["schedule_type"]).strip()
+    if "cron_expr" in body:
+        updates["cron_expr"] = str(body["cron_expr"]).strip()[:256]
+    if "at_time" in body:
+        updates["at_time"] = str(body["at_time"]).strip()[:64]
+    if "target_mode" in body:
+        updates["target_mode"] = str(body["target_mode"]).strip()[:256]
+    if "model" in body:
+        updates["model"] = str(body["model"]).strip()[:128]
+    if "initial_message" in body:
+        updates["initial_message"] = str(body["initial_message"]).strip()[:4096]
+    if "auto_approve" in body:
+        updates["auto_approve"] = bool(body["auto_approve"])
+    if "auto_approve_tools" in body:
+        raw = body["auto_approve_tools"]
+        updates["auto_approve_tools"] = raw if isinstance(raw, list) else []
+    if "enabled" in body:
+        updates["enabled"] = bool(body["enabled"])
+
+    # Validate schedule fields if changed
+    stype = updates.get("schedule_type", existing["schedule_type"])
+    cexpr = updates.get("cron_expr", existing["cron_expr"])
+    atime = updates.get("at_time", existing["at_time"])
+    schedule_fields_changed = (
+        "schedule_type" in updates or "cron_expr" in updates or "at_time" in updates
+    )
+    if schedule_fields_changed:
+        validation_err = _validate_schedule_fields(stype, cexpr, atime)
+        if validation_err:
+            return JSONResponse({"error": validation_err}, status_code=400)
+
+    # Recompute next_run if schedule changed or enabled toggled
+    if schedule_fields_changed or "enabled" in updates:
+        enabled = updates.get("enabled", bool(existing.get("enabled", 1)))
+        if enabled:
+            # Re-validate at_time when re-enabling a one-shot task
+            if stype == "at" and not schedule_fields_changed:
+                validation_err = _validate_schedule_fields(stype, cexpr, atime)
+                if validation_err:
+                    return JSONResponse({"error": validation_err}, status_code=400)
+            updates["next_run"] = _compute_next_run(stype, cexpr, atime)
+        else:
+            updates["next_run"] = ""
+
+    storage.update_scheduled_task(task_id, **updates)
+    task = storage.get_scheduled_task(task_id)
+    if task:
+        _normalize_task_dict(task)
+    return JSONResponse(task)
+
+
+async def admin_delete_schedule(request: Request) -> JSONResponse:
+    """DELETE /v1/api/admin/schedules/{task_id} — delete task + runs."""
+    from turnstone.core.web_helpers import require_storage_or_503
+
+    storage, err = require_storage_or_503(request)
+    if err:
+        return err
+    task_id = request.path_params["task_id"]
+    if storage.delete_scheduled_task(task_id):
+        return JSONResponse({"status": "ok"})
+    return JSONResponse({"error": "Schedule not found"}, status_code=404)
+
+
+async def admin_list_schedule_runs(request: Request) -> JSONResponse:
+    """GET /v1/api/admin/schedules/{task_id}/runs — run history."""
+    from turnstone.core.web_helpers import require_storage_or_503
+
+    storage, err = require_storage_or_503(request)
+    if err:
+        return err
+    task_id = request.path_params["task_id"]
+
+    # Verify task exists
+    if storage.get_scheduled_task(task_id) is None:
+        return JSONResponse({"error": "Schedule not found"}, status_code=404)
+
+    try:
+        limit = min(int(request.query_params.get("limit", "50")), 200)
+    except (ValueError, TypeError):
+        limit = 50
+    runs = storage.list_task_runs(task_id, limit=limit)
+    return JSONResponse({"runs": runs})
+
+
 # ---------------------------------------------------------------------------
 # App factory
 # ---------------------------------------------------------------------------
@@ -937,6 +1214,16 @@ def create_app(
                        admin_delete_channel,
                        methods=["DELETE"],
                    ),
+                    Route("/api/admin/schedules", admin_list_schedules),
+                    Route("/api/admin/schedules", admin_create_schedule, methods=["POST"]),
+                    Route("/api/admin/schedules/{task_id}", admin_get_schedule),
+                    Route("/api/admin/schedules/{task_id}", admin_update_schedule, methods=["PUT"]),
+                    Route(
+                        "/api/admin/schedules/{task_id}",
+                        admin_delete_schedule,
+                        methods=["DELETE"],
+                    ),
+                    Route("/api/admin/schedules/{task_id}/runs", admin_list_schedule_runs),
                ],
            ),
            Route("/health", health),
@@ -966,6 +1253,20 @@ def create_app(
    from turnstone.core.auth import LoginRateLimiter

    app.state.login_limiter = LoginRateLimiter()
+
+    # Scheduler — start background thread if storage is available
+    if auth_storage is not None:
+        from turnstone.console.scheduler import TaskScheduler
+
+        scheduler = TaskScheduler(
+            broker=broker,
+            collector=collector,
+            storage=auth_storage,
+        )
+        app.state.scheduler = scheduler
+    else:
+        app.state.scheduler = None
+
    return app


@@ -1100,7 +1401,7 @@ def main() -> None:

        proxy_token_mgr = ServiceTokenManager(
            user_id="console-proxy",
-            scopes=frozenset({"write"}),
+            scopes=frozenset({"read", "write", "approve"}),
            source="console",
            secret=jwt_secret,
            audience=JWT_AUD_SERVER,
@@ -46,10 +46,13 @@ function switchAdminTab(tab) {
    tab === "tokens" ? "" : "none";
  document.getElementById("admin-channels").style.display =
    tab === "channels" ? "" : "none";
+  document.getElementById("admin-schedules").style.display =
+    tab === "schedules" ? "" : "none";

  if (tab === "users") loadAdminUsers();
  if (tab === "tokens") _populateTokenUserSelect();
  if (tab === "channels") _populateChannelUserSelect();
+  if (tab === "schedules") loadAdminSchedules();
 }

 // ---------------------------------------------------------------------------
@@ -373,6 +376,517 @@ function confirmUnlinkChannel(channelType, channelUserId) {
  );
 }

+// ---------------------------------------------------------------------------
+// Schedules
+// ---------------------------------------------------------------------------
+
+var _csTrapHandler = null;
+var _esTrapHandler = null;
+var _srTrapHandler = null;
+var _editScheduleTriggerEl = null;
+var _runsScheduleTriggerEl = null;
+
+function loadAdminSchedules() {
+  authFetch("/v1/api/admin/schedules")
+    .then(function (r) {
+      if (!r.ok) throw new Error("Failed to load schedules");
+      return r.json();
+    })
+    .then(function (data) {
+      _renderSchedules(data.schedules || []);
+    })
+    .catch(function () {
+      document.getElementById("admin-schedules-table").innerHTML =
+        '<div class="dashboard-empty">Failed to load schedules</div>';
+    });
+}
+
+function _renderSchedules(schedules) {
+  var container = document.getElementById("admin-schedules-table");
+  if (!schedules.length) {
+    container.innerHTML =
+      '<div class="dashboard-empty">No scheduled tasks. Create one to get started.</div>';
+    return;
+  }
+  var html = "";
+  for (var i = 0; i < schedules.length; i++) {
+    var s = schedules[i];
+    var typeLabel = s.schedule_type === "cron" ? "cron" : "at";
+    var typeCls = s.schedule_type === "cron" ? "scope-write" : "scope-approve";
+    var schedule =
+      s.schedule_type === "cron"
+        ? s.cron_expr
+        : (s.at_time || "").slice(0, 16).replace("T", " ");
+    var target = s.target_mode;
+    var nextRun = s.next_run
+      ? escapeHtml(s.next_run).slice(0, 16).replace("T", " ")
+      : "\u2014";
+    var enabled = s.enabled;
+    var statusCls = enabled ? "sched-active" : "sched-disabled";
+    var statusLabel = enabled ? "active" : "disabled";
+    var statusDot = enabled ? "\u25cf " : "\u25cb ";
+    if (s.schedule_type === "at" && !enabled && s.last_run) {
+      statusCls = "sched-expired";
+      statusLabel = "completed";
+      statusDot = "\u25c9 ";
+    }
+    html +=
+      '<div class="admin-row" role="listitem">' +
+      '<span class="admin-col admin-col-sname">' +
+      escapeHtml(s.name) +
+      "</span>" +
+      '<span class="admin-col admin-col-stype"><span class="scope-badge ' +
+      typeCls +
+      '">' +
+      typeLabel +
+      "</span></span>" +
+      '<span class="admin-col admin-col-sschedule"><code>' +
+      escapeHtml(schedule) +
+      "</code></span>" +
+      '<span class="admin-col admin-col-starget">' +
+      escapeHtml(target) +
+      "</span>" +
+      '<span class="admin-col admin-col-snext">' +
+      nextRun +
+      "</span>" +
+      '<span class="admin-col admin-col-sstatus"><span class="' +
+      statusCls +
+      '">' +
+      statusDot +
+      statusLabel +
+      "</span></span>" +
+      '<span class="admin-col admin-col-actions">' +
+      '<button class="admin-btn-action" data-edit-sched="' +
+      escapeHtml(s.task_id) +
+      '" title="Edit">edit</button>' +
+      '<button class="admin-btn-action" data-runs-sched="' +
+      escapeHtml(s.task_id) +
+      '" title="Run history">runs</button>' +
+      '<button class="admin-btn-action" data-toggle-sched="' +
+      escapeHtml(s.task_id) +
+      '" data-enabled="' +
+      (enabled ? "1" : "0") +
+      '" title="' +
+      (enabled ? "Disable" : "Enable") +
+      '">' +
+      (enabled ? "disable" : "enable") +
+      "</button>" +
+      '<button class="admin-btn-danger" data-delete-sched="' +
+      escapeHtml(s.task_id) +
+      '" data-sname="' +
+      escapeHtml(s.name) +
+      '" title="Delete">delete</button>' +
+      "</span></div>";
+  }
+  container.innerHTML = html;
+  // Bind buttons
+  var editBtns = container.querySelectorAll("[data-edit-sched]");
+  for (var j = 0; j < editBtns.length; j++) {
+    editBtns[j].addEventListener("click", function () {
+      showEditScheduleModal(this.getAttribute("data-edit-sched"));
+    });
+  }
+  var runsBtns = container.querySelectorAll("[data-runs-sched]");
+  for (var k = 0; k < runsBtns.length; k++) {
+    runsBtns[k].addEventListener("click", function () {
+      showScheduleRuns(this.getAttribute("data-runs-sched"));
+    });
+  }
+  var toggleBtns = container.querySelectorAll("[data-toggle-sched]");
+  for (var m = 0; m < toggleBtns.length; m++) {
+    toggleBtns[m].addEventListener("click", function () {
+      toggleSchedule(
+        this.getAttribute("data-toggle-sched"),
+        this.getAttribute("data-enabled") === "1",
+      );
+    });
+  }
+  var delBtns = container.querySelectorAll("[data-delete-sched]");
+  for (var n = 0; n < delBtns.length; n++) {
+    delBtns[n].addEventListener("click", function () {
+      confirmDeleteSchedule(
+        this.getAttribute("data-delete-sched"),
+        this.getAttribute("data-sname"),
+      );
+    });
+  }
+}
+
+function toggleSchedule(taskId, currentlyEnabled) {
+  authFetch("/v1/api/admin/schedules/" + encodeURIComponent(taskId), {
+    method: "PUT",
+    headers: { "Content-Type": "application/json" },
+    body: JSON.stringify({ enabled: !currentlyEnabled }),
+  })
+    .then(function (r) {
+      if (!r.ok) throw new Error("Toggle failed");
+      showToast(currentlyEnabled ? "Schedule disabled" : "Schedule enabled");
+      loadAdminSchedules();
+    })
+    .catch(function () {
+      showToast("Failed to toggle schedule");
+    });
+}
+
+function confirmDeleteSchedule(taskId, name) {
+  showConfirmModal(
+    "Delete Schedule",
+    "Delete schedule \u2018" +
+      name +
+      "\u2019 and its run history? This cannot be undone.",
+    "Delete",
+    function () {
+      authFetch("/v1/api/admin/schedules/" + encodeURIComponent(taskId), {
+        method: "DELETE",
+      })
+        .then(function (r) {
+          if (!r.ok) throw new Error("Delete failed");
+          showToast("Schedule deleted");
+          loadAdminSchedules();
+        })
+        .catch(function () {
+          showToast("Failed to delete schedule");
+        });
+    },
+  );
+}
+
+// --- Create Schedule Modal ---
+
+function toggleScheduleTypeFields() {
+  var t = document.getElementById("cs-type").value;
+  document.getElementById("cs-cron-group").style.display =
+    t === "cron" ? "" : "none";
+  document.getElementById("cs-at-group").style.display =
+    t === "at" ? "" : "none";
+  if (t === "cron") document.getElementById("cs-cron").focus();
+  else document.getElementById("cs-at").focus();
+}
+
+function toggleScheduleNodeField() {
+  var v = document.getElementById("cs-target").value;
+  document.getElementById("cs-node-group").style.display =
+    v === "node" ? "" : "none";
+  if (v === "node") document.getElementById("cs-node").focus();
+}
+
+function showCreateScheduleModal() {
+  var overlay = document.getElementById("create-schedule-overlay");
+  overlay.style.display = "flex";
+  document.getElementById("create-schedule-error").style.display = "none";
+  document.getElementById("cs-name").value = "";
+  document.getElementById("cs-desc").value = "";
+  document.getElementById("cs-type").value = "cron";
+  document.getElementById("cs-cron").value = "";
+  document.getElementById("cs-at").value = "";
+  document.getElementById("cs-target").value = "auto";
+  document.getElementById("cs-node").value = "";
+  document.getElementById("cs-model").value = "";
+  document.getElementById("cs-message").value = "";
+  document.getElementById("cs-autoapprove").checked = false;
+  toggleScheduleTypeFields();
+  toggleScheduleNodeField();
+  document.getElementById("cs-submit").disabled = false;
+  document.getElementById("cs-submit").textContent = "Create";
+  _csTrapHandler = _installTrap(
+    "create-schedule-overlay",
+    "create-schedule-box",
+  );
+  setTimeout(function () {
+    document.getElementById("cs-name").focus();
+  }, 50);
+}
+
+function hideCreateScheduleModal() {
+  document.getElementById("create-schedule-overlay").style.display = "none";
+  _csTrapHandler = _removeTrap(_csTrapHandler);
+  var trigger = document.querySelector("#admin-schedules .admin-action-btn");
+  if (trigger) trigger.focus();
+}
+
+function submitCreateSchedule() {
+  var name = (document.getElementById("cs-name").value || "").trim();
+  var desc = (document.getElementById("cs-desc").value || "").trim();
+  var schedType = document.getElementById("cs-type").value;
+  var cronExpr = (document.getElementById("cs-cron").value || "").trim();
+  var atTime = document.getElementById("cs-at").value || "";
+  var targetMode = document.getElementById("cs-target").value;
+  var nodeId = (document.getElementById("cs-node").value || "").trim();
+  var model = (document.getElementById("cs-model").value || "").trim();
+  var message = (document.getElementById("cs-message").value || "").trim();
+  var autoApprove = document.getElementById("cs-autoapprove").checked;
+  var errEl = document.getElementById("create-schedule-error");
+
+  if (!name) return _showModalError(errEl, "Name is required");
+  if (!message) return _showModalError(errEl, "Initial message is required");
+  if (schedType === "cron" && !cronExpr)
+    return _showModalError(errEl, "Cron expression is required");
+  if (schedType === "at" && !atTime)
+    return _showModalError(errEl, "Run time is required");
+
+  // Normalize datetime-local to "YYYY-MM-DDTHH:MM:SS+00:00" (UTC)
+  if (schedType === "at" && atTime) {
+    if (atTime.length === 16) atTime += ":00";
+    else if (atTime.length > 19) atTime = atTime.slice(0, 19);
+    atTime += "+00:00";
+  }
+
+  if (targetMode === "node") targetMode = nodeId;
+
+  var btn = document.getElementById("cs-submit");
+  btn.disabled = true;
+  btn.textContent = "Creating\u2026";
+
+  authFetch("/v1/api/admin/schedules", {
+    method: "POST",
+    headers: { "Content-Type": "application/json" },
+    body: JSON.stringify({
+      name: name,
+      description: desc,
+      schedule_type: schedType,
+      cron_expr: cronExpr,
+      at_time: atTime,
+      target_mode: targetMode,
+      model: model,
+      initial_message: message,
+      auto_approve: autoApprove,
+    }),
+  })
+    .then(function (r) {
+      if (!r.ok)
+        return r.json().then(function (d) {
+          throw new Error(d.error || "Failed");
+        });
+      return r.json();
+    })
+    .then(function () {
+      hideCreateScheduleModal();
+      showToast("Schedule '" + name + "' created");
+      loadAdminSchedules();
+    })
+    .catch(function (err) {
+      btn.disabled = false;
+      btn.textContent = "Create";
+      _showModalError(errEl, err.message || "Failed to create schedule");
+    });
+}
+
+// --- Edit Schedule Modal ---
+
+function toggleEditScheduleTypeFields() {
+  var t = document.getElementById("es-type").value;
+  document.getElementById("es-cron-group").style.display =
+    t === "cron" ? "" : "none";
+  document.getElementById("es-at-group").style.display =
+    t === "at" ? "" : "none";
+  if (t === "cron") document.getElementById("es-cron").focus();
+  else document.getElementById("es-at").focus();
+}
+
+function toggleEditScheduleNodeField() {
+  var v = document.getElementById("es-target").value;
+  document.getElementById("es-node-group").style.display =
+    v === "node" ? "" : "none";
+  if (v === "node") document.getElementById("es-node").focus();
+}
+
+function showEditScheduleModal(taskId) {
+  _editScheduleTriggerEl = document.activeElement;
+  authFetch("/v1/api/admin/schedules/" + encodeURIComponent(taskId))
+    .then(function (r) {
+      if (!r.ok) throw new Error("Not found");
+      return r.json();
+    })
+    .then(function (s) {
+      document.getElementById("es-id").value = s.task_id;
+      document.getElementById("es-name").value = s.name || "";
+      document.getElementById("es-desc").value = s.description || "";
+      document.getElementById("es-type").value = s.schedule_type;
+      document.getElementById("es-cron").value = s.cron_expr || "";
+      document.getElementById("es-at").value = (s.at_time || "").slice(0, 16);
+      var isSpecificNode =
+        s.target_mode &&
+        s.target_mode !== "auto" &&
+        s.target_mode !== "pool" &&
+        s.target_mode !== "all";
+      document.getElementById("es-target").value = isSpecificNode
+        ? "node"
+        : s.target_mode;
+      document.getElementById("es-node").value = isSpecificNode
+        ? s.target_mode
+        : "";
+      document.getElementById("es-model").value = s.model || "";
+      document.getElementById("es-message").value = s.initial_message || "";
+      document.getElementById("es-autoapprove").checked = !!s.auto_approve;
+      document.getElementById("es-enabled").checked = !!s.enabled;
+      toggleEditScheduleTypeFields();
+      toggleEditScheduleNodeField();
+      document.getElementById("edit-schedule-error").style.display = "none";
+      document.getElementById("es-submit").disabled = false;
+      document.getElementById("es-submit").textContent = "Save";
+      var overlay = document.getElementById("edit-schedule-overlay");
+      overlay.style.display = "flex";
+      _esTrapHandler = _installTrap(
+        "edit-schedule-overlay",
+        "edit-schedule-box",
+      );
+      setTimeout(function () {
+        document.getElementById("es-name").focus();
+      }, 50);
+    })
+    .catch(function () {
+      showToast("Failed to load schedule");
+    });
+}
+
+function hideEditScheduleModal() {
+  document.getElementById("edit-schedule-overlay").style.display = "none";
+  _esTrapHandler = _removeTrap(_esTrapHandler);
+  if (_editScheduleTriggerEl && _editScheduleTriggerEl.isConnected) {
+    _editScheduleTriggerEl.focus();
+  }
+  _editScheduleTriggerEl = null;
+}
+
+function submitEditSchedule() {
+  var taskId = document.getElementById("es-id").value;
+  var name = (document.getElementById("es-name").value || "").trim();
+  var message = (document.getElementById("es-message").value || "").trim();
+  var schedType = document.getElementById("es-type").value;
+  var cronExpr = (document.getElementById("es-cron").value || "").trim();
+  var targetMode = document.getElementById("es-target").value;
+  if (targetMode === "node")
+    targetMode = (document.getElementById("es-node").value || "").trim();
+  var atTime = document.getElementById("es-at").value || "";
+  if (atTime) {
+    if (atTime.length === 16) atTime += ":00";
+    else if (atTime.length > 19) atTime = atTime.slice(0, 19);
+    atTime += "+00:00";
+  }
+
+  var errEl = document.getElementById("edit-schedule-error");
+
+  if (!name) return _showModalError(errEl, "Name is required");
+  if (!message) return _showModalError(errEl, "Initial message is required");
+  if (schedType === "cron" && !cronExpr)
+    return _showModalError(errEl, "Cron expression is required");
+  if (schedType === "at" && !atTime)
+    return _showModalError(errEl, "Run time is required");
+
+  var btn = document.getElementById("es-submit");
+  btn.disabled = true;
+  btn.textContent = "Saving\u2026";
+
+  authFetch("/v1/api/admin/schedules/" + encodeURIComponent(taskId), {
+    method: "PUT",
+    headers: { "Content-Type": "application/json" },
+    body: JSON.stringify({
+      name: (document.getElementById("es-name").value || "").trim(),
+      description: (document.getElementById("es-desc").value || "").trim(),
+      schedule_type: document.getElementById("es-type").value,
+      cron_expr: (document.getElementById("es-cron").value || "").trim(),
+      at_time: atTime,
+      target_mode: targetMode,
+      model: (document.getElementById("es-model").value || "").trim(),
+      initial_message: (
+        document.getElementById("es-message").value || ""
+      ).trim(),
+      auto_approve: document.getElementById("es-autoapprove").checked,
+      enabled: document.getElementById("es-enabled").checked,
+    }),
+  })
+    .then(function (r) {
+      if (!r.ok)
+        return r.json().then(function (d) {
+          throw new Error(d.error || "Failed");
+        });
+      return r.json();
+    })
+    .then(function () {
+      hideEditScheduleModal();
+      showToast("Schedule updated");
+      loadAdminSchedules();
+    })
+    .catch(function (err) {
+      btn.disabled = false;
+      btn.textContent = "Save";
+      _showModalError(errEl, err.message || "Failed to update schedule");
+    });
+}
+
+// --- Schedule Runs Modal ---
+
+function showScheduleRuns(taskId) {
+  _runsScheduleTriggerEl = document.activeElement;
+  authFetch(
+    "/v1/api/admin/schedules/" + encodeURIComponent(taskId) + "/runs?limit=50",
+  )
+    .then(function (r) {
+      if (!r.ok) throw new Error("Not found");
+      return r.json();
+    })
+    .then(function (data) {
+      var runs = data.runs || [];
+      var container = document.getElementById("schedule-runs-table");
+      if (!runs.length) {
+        container.innerHTML = '<div class="dashboard-empty">No runs yet</div>';
+      } else {
+        var html =
+          '<div class="admin-colheaders sched-runs-grid" aria-hidden="true">' +
+          '<span class="admin-col">STARTED</span>' +
+          '<span class="admin-col">NODE</span>' +
+          '<span class="admin-col">STATUS</span>' +
+          '<span class="admin-col">ERROR</span></div>';
+        for (var i = 0; i < runs.length; i++) {
+          var r = runs[i];
+          var statusCls =
+            r.status === "dispatched"
+              ? "sched-active"
+              : r.status === "failed"
+                ? "sched-expired"
+                : "";
+          html +=
+            '<div class="admin-row sched-runs-grid">' +
+            '<span class="admin-col">' +
+            escapeHtml(r.started || "")
+              .slice(0, 19)
+              .replace("T", " ") +
+            "</span>" +
+            '<span class="admin-col">' +
+            escapeHtml(r.node_id || "\u2014") +
+            "</span>" +
+            '<span class="admin-col"><span class="' +
+            statusCls +
+            '">' +
+            escapeHtml(r.status) +
+            "</span></span>" +
+            '<span class="admin-col">' +
+            escapeHtml(r.error || "\u2014") +
+            "</span></div>";
+        }
+        container.innerHTML = html;
+      }
+      var overlay = document.getElementById("schedule-runs-overlay");
+      overlay.style.display = "flex";
+      _srTrapHandler = _installTrap(
+        "schedule-runs-overlay",
+        "schedule-runs-box",
+      );
+    })
+    .catch(function () {
+      showToast("Failed to load run history");
+    });
+}
+
+function hideScheduleRunsModal() {
+  document.getElementById("schedule-runs-overlay").style.display = "none";
+  _srTrapHandler = _removeTrap(_srTrapHandler);
+  if (_runsScheduleTriggerEl && _runsScheduleTriggerEl.isConnected) {
+    _runsScheduleTriggerEl.focus();
+  }
+  _runsScheduleTriggerEl = null;
+}
+
 // ---------------------------------------------------------------------------
 // Create Channel Link Modal
 // ---------------------------------------------------------------------------
@@ -644,7 +1158,7 @@ function _modalFocusTrap(boxId) {
      var box = document.getElementById(boxId);
      if (!box) return;
      var focusable = box.querySelectorAll(
-        "input:not([disabled]), select:not([disabled]), button:not([disabled])",
+        "input:not([disabled]):not([type='hidden']), select:not([disabled]), textarea:not([disabled]), button:not([disabled])",
      );
      var visible = [];
      for (var i = 0; i < focusable.length; i++) {
@@ -678,6 +1192,10 @@ function _installTrap(overlayId, boxId, trapRef) {
        else if (overlayId === "token-created-overlay") hideTokenCreatedModal();
        else if (overlayId === "create-channel-overlay")
          hideCreateChannelModal();
+        else if (overlayId === "create-schedule-overlay")
+          hideCreateScheduleModal();
+        else if (overlayId === "edit-schedule-overlay") hideEditScheduleModal();
+        else if (overlayId === "schedule-runs-overlay") hideScheduleRunsModal();
        else if (overlayId === "confirm-overlay") hideConfirmModal();
      }
    };
@@ -721,6 +1239,24 @@ document.addEventListener("keydown", function (e) {
    hideCreateChannelModal();
    return;
  }
+  var cso = document.getElementById("create-schedule-overlay");
+  if (cso && cso.style.display !== "none") {
+    e.preventDefault();
+    hideCreateScheduleModal();
+    return;
+  }
+  var eso = document.getElementById("edit-schedule-overlay");
+  if (eso && eso.style.display !== "none") {
+    e.preventDefault();
+    hideEditScheduleModal();
+    return;
+  }
+  var sro = document.getElementById("schedule-runs-overlay");
+  if (sro && sro.style.display !== "none") {
+    e.preventDefault();
+    hideScheduleRunsModal();
+    return;
+  }
  var cf = document.getElementById("confirm-overlay");
  if (cf && cf.style.display !== "none") {
    e.preventDefault();
@@ -735,7 +1271,7 @@ document.addEventListener("keydown", function (e) {
  if (!tablist) return;
  tablist.addEventListener("keydown", function (e) {
    if (e.key !== "ArrowLeft" && e.key !== "ArrowRight") return;
-    var tabOrder = ["users", "tokens", "channels"];
+    var tabOrder = ["users", "tokens", "channels", "schedules"];
    var idx = tabOrder.indexOf(_adminTab);
    if (e.key === "ArrowRight") idx = (idx + 1) % tabOrder.length;
    else idx = (idx - 1 + tabOrder.length) % tabOrder.length;
@@ -80,6 +80,7 @@
      <button id="tab-users" class="admin-tab active" data-tab="users" role="tab" aria-selected="true" aria-controls="admin-users" tabindex="0" onclick="switchAdminTab('users')">Users</button>
      <button id="tab-tokens" class="admin-tab" data-tab="tokens" role="tab" aria-selected="false" aria-controls="admin-tokens" tabindex="-1" onclick="switchAdminTab('tokens')">Tokens</button>
      <button id="tab-channels" class="admin-tab" data-tab="channels" role="tab" aria-selected="false" aria-controls="admin-channels" tabindex="-1" onclick="switchAdminTab('channels')">Channels</button>
+      <button id="tab-schedules" class="admin-tab" data-tab="schedules" role="tab" aria-selected="false" aria-controls="admin-schedules" tabindex="-1" onclick="switchAdminTab('schedules')">Schedules</button>
    </div>

    <!-- Users Tab -->
@@ -142,6 +143,26 @@
        <div class="dashboard-empty">Select a user to view channel links</div>
      </div>
    </div>
+
+    <!-- Schedules Tab -->
+    <div id="admin-schedules" class="admin-panel" role="tabpanel" aria-labelledby="tab-schedules" style="display:none">
+      <div class="admin-toolbar">
+        <span class="section-header" style="margin:0">SCHEDULED TASKS</span>
+        <button class="admin-action-btn" onclick="showCreateScheduleModal()">+ New schedule</button>
+      </div>
+      <div class="admin-colheaders" aria-hidden="true">
+        <span class="admin-col admin-col-sname">NAME</span>
+        <span class="admin-col admin-col-stype">TYPE</span>
+        <span class="admin-col admin-col-sschedule">SCHEDULE</span>
+        <span class="admin-col admin-col-starget">TARGET</span>
+        <span class="admin-col admin-col-snext">NEXT RUN</span>
+        <span class="admin-col admin-col-sstatus">STATUS</span>
+        <span class="admin-col admin-col-actions">ACTIONS</span>
+      </div>
+      <div id="admin-schedules-table" role="list" aria-label="Scheduled tasks" aria-live="polite">
+        <div class="dashboard-empty">Loading schedules...</div>
+      </div>
+    </div>
  </div>
 </div>

@@ -284,6 +305,110 @@ window.TURNSTONE_KB_SHORTCUTS = [
  </div>
 </div>

+<!-- Create Schedule Modal -->
+<div id="create-schedule-overlay" style="display:none" role="dialog" aria-modal="true" aria-labelledby="create-schedule-title">
+  <div id="create-schedule-box" class="admin-modal admin-modal-wide">
+    <h2 id="create-schedule-title">New Schedule</h2>
+    <div id="create-schedule-error" role="alert" aria-live="assertive"></div>
+    <label for="cs-name">Name</label>
+    <input id="cs-name" type="text" placeholder="Daily health check" autocomplete="off">
+    <label for="cs-desc">Description <span class="label-hint">optional</span></label>
+    <input id="cs-desc" type="text" placeholder="" autocomplete="off">
+    <label for="cs-type">Schedule type</label>
+    <select id="cs-type" onchange="toggleScheduleTypeFields()">
+      <option value="cron">Cron (recurring)</option>
+      <option value="at">At (one-shot)</option>
+    </select>
+    <div id="cs-cron-group">
+      <label for="cs-cron">Cron expression</label>
+      <input id="cs-cron" type="text" placeholder="0 9 * * MON-FRI" autocomplete="off" spellcheck="false" aria-describedby="cs-cron-hint">
+      <span id="cs-cron-hint" class="label-hint" style="display:block;margin-top:3px">min hour day month weekday</span>
+    </div>
+    <div id="cs-at-group" style="display:none">
+      <label for="cs-at">Run at</label>
+      <input id="cs-at" type="datetime-local">
+    </div>
+    <label for="cs-target">Target</label>
+    <select id="cs-target" onchange="toggleScheduleNodeField()">
+      <option value="auto">Auto (best available)</option>
+      <option value="pool">Pool (any bridge)</option>
+      <option value="all">All nodes</option>
+      <option value="node">Specific node...</option>
+    </select>
+    <div id="cs-node-group" style="display:none">
+      <label for="cs-node">Node ID</label>
+      <input id="cs-node" type="text" placeholder="node-001" autocomplete="off" spellcheck="false">
+    </div>
+    <label for="cs-model">Model <span class="label-hint">optional</span></label>
+    <input id="cs-model" type="text" placeholder="Default model" autocomplete="off">
+    <label for="cs-message">Initial message</label>
+    <textarea id="cs-message" rows="3" placeholder="What should the workstream do?"></textarea>
+    <label class="admin-checkbox"><input id="cs-autoapprove" type="checkbox"> Auto-approve tool calls</label>
+    <div class="modal-buttons">
+      <button class="modal-cancel" onclick="hideCreateScheduleModal()">Cancel</button>
+      <button id="cs-submit" class="modal-submit" onclick="submitCreateSchedule()">Create</button>
+    </div>
+  </div>
+</div>
+
+<!-- Edit Schedule Modal -->
+<div id="edit-schedule-overlay" style="display:none" role="dialog" aria-modal="true" aria-labelledby="edit-schedule-title">
+  <div id="edit-schedule-box" class="admin-modal admin-modal-wide">
+    <h2 id="edit-schedule-title">Edit Schedule</h2>
+    <div id="edit-schedule-error" role="alert" aria-live="assertive"></div>
+    <input id="es-id" type="hidden">
+    <label for="es-name">Name</label>
+    <input id="es-name" type="text" autocomplete="off">
+    <label for="es-desc">Description</label>
+    <input id="es-desc" type="text" autocomplete="off">
+    <label for="es-type">Schedule type</label>
+    <select id="es-type" onchange="toggleEditScheduleTypeFields()">
+      <option value="cron">Cron (recurring)</option>
+      <option value="at">At (one-shot)</option>
+    </select>
+    <div id="es-cron-group">
+      <label for="es-cron">Cron expression</label>
+      <input id="es-cron" type="text" autocomplete="off" spellcheck="false">
+    </div>
+    <div id="es-at-group" style="display:none">
+      <label for="es-at">Run at</label>
+      <input id="es-at" type="datetime-local">
+    </div>
+    <label for="es-target">Target</label>
+    <select id="es-target" onchange="toggleEditScheduleNodeField()">
+      <option value="auto">Auto (best available)</option>
+      <option value="pool">Pool (any bridge)</option>
+      <option value="all">All nodes</option>
+      <option value="node">Specific node...</option>
+    </select>
+    <div id="es-node-group" style="display:none">
+      <label for="es-node">Node ID</label>
+      <input id="es-node" type="text" autocomplete="off" spellcheck="false">
+    </div>
+    <label for="es-model">Model</label>
+    <input id="es-model" type="text" autocomplete="off">
+    <label for="es-message">Initial message</label>
+    <textarea id="es-message" rows="3"></textarea>
+    <label class="admin-checkbox"><input id="es-autoapprove" type="checkbox"> Auto-approve tool calls</label>
+    <label class="admin-checkbox"><input id="es-enabled" type="checkbox"> Enabled</label>
+    <div class="modal-buttons">
+      <button class="modal-cancel" onclick="hideEditScheduleModal()">Cancel</button>
+      <button id="es-submit" class="modal-submit" onclick="submitEditSchedule()">Save</button>
+    </div>
+  </div>
+</div>
+
+<!-- Schedule Runs Modal -->
+<div id="schedule-runs-overlay" style="display:none" role="dialog" aria-modal="true" aria-labelledby="schedule-runs-title">
+  <div id="schedule-runs-box" class="admin-modal admin-modal-wide">
+    <h2 id="schedule-runs-title">Run History</h2>
+    <div id="schedule-runs-table"></div>
+    <div class="modal-buttons">
+      <button class="modal-cancel" onclick="hideScheduleRunsModal()">Close</button>
+    </div>
+  </div>
+</div>
+
 <script src="/static/admin.js"></script>
 <script src="/static/app.js"></script>
 </body>
@@ -832,6 +832,58 @@
 .admin-btn-danger:hover { opacity: 1; background: rgba(248, 113, 113, 0.1); }
 .admin-btn-danger:focus-visible { outline: 2px solid var(--red); outline-offset: 2px; }

+.admin-btn-action {
+  background: none;
+  border: 1px solid var(--border-strong);
+  color: var(--fg-dim);
+  font-family: var(--font-display);
+  font-size: 10px;
+  font-weight: 500;
+  padding: 2px 8px;
+  border-radius: var(--radius-sm);
+  cursor: pointer;
+  opacity: 0.8;
+  transition: opacity 0.15s, background 0.15s;
+  margin-right: 4px;
+}
+.admin-btn-action:hover { opacity: 1; background: rgba(255, 255, 255, 0.05); color: var(--fg); }
+.admin-btn-action:focus-visible { outline: 2px solid var(--accent); outline-offset: 2px; }
+
+/* Schedules grid: NAME | TYPE | SCHEDULE | TARGET | NEXT RUN | STATUS | ACTIONS */
+#admin-schedules .admin-colheaders,
+#admin-schedules .admin-row {
+  grid-template-columns: 1.5fr 60px 1.2fr 80px 130px 70px 170px;
+}
+
+/* Schedule runs grid: STARTED | NODE | STATUS | ERROR */
+.sched-runs-grid { grid-template-columns: 2fr 1fr 1fr 2fr; }
+
+/* Schedule status indicators */
+.sched-active { color: var(--green); font-weight: 500; }
+.sched-disabled { color: var(--fg-dim); }
+.sched-expired { color: var(--accent); }
+
+/* Wide modal variant for schedule forms */
+.admin-modal-wide { width: 480px; }
+
+/* Checkbox labels inside admin modals */
+.admin-modal label.admin-checkbox {
+  display: flex;
+  align-items: center;
+  gap: 8px;
+  font-size: 12px;
+  font-weight: 500;
+  text-transform: none;
+  letter-spacing: 0;
+  color: var(--fg);
+  cursor: pointer;
+  margin-top: 14px;
+}
+.admin-modal label.admin-checkbox input[type="checkbox"] {
+  width: auto;
+  margin: 0;
+}
+
 /* Admin modals (reuse new-ws-overlay pattern) */
 .admin-modal {
  background: var(--bg-surface);
@@ -871,7 +923,7 @@
  margin-top: 12px;
 }
 .admin-modal label:first-of-type { margin-top: 0; }
-.admin-modal input, .admin-modal select {
+.admin-modal input:not([type="hidden"]), .admin-modal select, .admin-modal textarea {
  width: 100%;
  padding: 9px 12px;
  background: var(--bg);
@@ -882,12 +934,13 @@
  font-size: 13px;
  transition: border-color 0.15s, box-shadow 0.15s;
 }
-.admin-modal input:focus, .admin-modal select:focus {
+.admin-modal input:focus, .admin-modal select:focus, .admin-modal textarea:focus {
  border-color: var(--accent);
  outline: none;
  box-shadow: 0 0 0 3px var(--accent-dim);
 }
-.admin-modal input::placeholder { color: var(--fg-dim); opacity: 0.6; }
+.admin-modal input::placeholder, .admin-modal textarea::placeholder { color: var(--fg-dim); opacity: 0.6; }
+.admin-modal textarea { resize: vertical; min-height: 40px; }
 .admin-modal [role="alert"] { color: var(--red); font-size: 12px; margin-bottom: 8px; display: none; }

 .modal-buttons { display: flex; gap: 10px; margin-top: 20px; }
@@ -925,7 +978,8 @@
 .modal-submit:focus-visible { outline: 2px solid var(--fg-bright); outline-offset: 2px; }
 .modal-submit:disabled { opacity: 0.4; cursor: not-allowed; filter: none; }

-#create-user-overlay, #create-token-overlay, #token-created-overlay, #create-channel-overlay, #confirm-overlay {
+#create-user-overlay, #create-token-overlay, #token-created-overlay, #create-channel-overlay, #confirm-overlay,
+#create-schedule-overlay, #edit-schedule-overlay, #schedule-runs-overlay {
  position: fixed;
  inset: 0;
  background: rgba(0, 0, 0, 0.7);
@@ -969,6 +1023,10 @@
    grid-template-columns: 80px 1fr 80px;
  }
  #admin-channels .admin-col-created { display: none; }
+  #admin-schedules .admin-colheaders, #admin-schedules .admin-row {
+    grid-template-columns: 1fr 60px 80px 130px;
+  }
+  .admin-col-sschedule, .admin-col-starget, .admin-col-snext { display: none; }
 }

 /* ==========================================================================
@@ -981,7 +1039,7 @@
  .node-link, .dash-cell-node, .pagination button { transition: none; }
  .dash-row.has-link::after, .node-group-header::before { transition: none; }
  #new-ws-box select, #new-ws-box input, #new-ws-buttons button { transition: none; }
-  .admin-tab, .admin-row, .admin-btn-danger { transition: none; }
+  .admin-tab, .admin-row, .admin-btn-danger, .admin-btn-action { transition: none; }
  .admin-action-btn, .modal-cancel, .modal-submit { transition: none; }
  .admin-modal input, .admin-modal select { transition: none; }
 }
@@ -49,6 +49,7 @@ TOKEN_BYTES = 32  # 64 hex chars after prefix
 JWT_ISSUER = "turnstone"
 JWT_AUD_SERVER = "turnstone-server"
 JWT_AUD_CONSOLE = "turnstone-console"
+JWT_AUD_CHANNEL = "turnstone-channel"
 _MIN_SECRET_LENGTH = 32  # 256 bits minimum for HMAC-SHA256

 VALID_SCOPES: frozenset[str] = frozenset({"read", "write", "approve"})
@@ -61,7 +61,7 @@ _CONFIG_MAP: dict[str, dict[str, str]] = {
    },
    "session": {
        "instructions": "instructions",
-        "retention_days": "session_retention_days",
+        "retention_days": "retention_days",
        "compact_max_tokens": "compact_max_tokens",
        "auto_compact_pct": "auto_compact_pct",
    },
@@ -21,22 +21,11 @@ def normalize_key(key: str) -> str:
    return key.lower().replace("-", "_").replace(" ", "_")


-# -- Core session operations ---------------------------------------------------
-
-
-def register_session(
-    session_id: str,
-    title: str | None = None,
-    node_id: str | None = None,
-    ws_id: str | None = None,
-) -> None:
-    """Create a sessions row for a new session (no-op if already exists)."""
-    with contextlib.suppress(Exception):
-        get_storage().register_session(session_id, title, node_id=node_id, ws_id=ws_id)
+# -- Core conversation operations ---------------------------------------------


 def save_message(
-    session_id: str,
+    ws_id: str,
    role: str,
    content: str | None,
    tool_name: str | None = None,
@@ -47,111 +36,19 @@ def save_message(
    """Log a message to the conversations table."""
    with contextlib.suppress(Exception):
        get_storage().save_message(
-            session_id, role, content, tool_name, tool_args, tool_call_id, provider_data
+            ws_id, role, content, tool_name, tool_args, tool_call_id, provider_data
        )


-def load_session_messages(session_id: str) -> list[dict[str, Any]]:
-    """Load messages for a session and reconstruct OpenAI message format."""
+def load_messages(ws_id: str) -> list[dict[str, Any]]:
+    """Load messages for a workstream and reconstruct OpenAI message format."""
    try:
-        return get_storage().load_session_messages(session_id)
+        return get_storage().load_messages(ws_id)
    except Exception:
        return []


-# -- Session management --------------------------------------------------------
-
-
-def list_sessions(limit: int = 20) -> list[Any]:
-    """List recent sessions with message counts."""
-    try:
-        return get_storage().list_sessions(limit)
-    except Exception:
-        return []
-
-
-def delete_session(session_id: str) -> bool:
-    """Delete a session and all its messages."""
-    try:
-        return get_storage().delete_session(session_id)
-    except Exception:
-        return False
-
-
-def prune_sessions(
-    retention_days: int = 90,
-    log_fn: Callable[[str], None] | None = None,
-) -> tuple[int, int]:
-    """Prune orphaned and stale sessions."""
-    try:
-        orphans, stale = get_storage().prune_sessions(retention_days)
-    except Exception:
-        return (0, 0)
-
-    if log_fn and (orphans or stale):
-        parts = []
-        if orphans:
-            parts.append(f"{orphans} empty session{'s' if orphans != 1 else ''}")
-        if stale:
-            parts.append(
-                f"{stale} session{'s' if stale != 1 else ''} older than {retention_days} days"
-            )
-        log_fn(f"[turnstone] Session cleanup: removed {', '.join(parts)}.")
-
-    return (orphans, stale)
-
-
-def resolve_session(alias_or_id: str) -> str | None:
-    """Resolve an alias or session_id (or prefix) to a full session_id."""
-    try:
-        return get_storage().resolve_session(alias_or_id)
-    except Exception:
-        return None
-
-
-# -- Session config ------------------------------------------------------------
-
-
-def save_session_config(session_id: str, config: dict[str, str]) -> None:
-    """Persist session configuration key/value pairs."""
-    with contextlib.suppress(Exception):
-        get_storage().save_session_config(session_id, config)
-
-
-def load_session_config(session_id: str) -> dict[str, str]:
-    """Load session configuration."""
-    try:
-        return get_storage().load_session_config(session_id)
-    except Exception:
-        return {}
-
-
-# -- Session metadata ----------------------------------------------------------
-
-
-def set_session_alias(session_id: str, alias: str) -> bool:
-    """Set a human-friendly alias. Returns False if alias is taken."""
-    try:
-        return get_storage().set_session_alias(session_id, alias)
-    except Exception:
-        return False
-
-
-def get_session_name(session_id: str) -> str | None:
-    """Return the alias (or title) for a session, or None if unset."""
-    try:
-        return get_storage().get_session_name(session_id)
-    except Exception:
-        return None
-
-
-def update_session_title(session_id: str, title: str) -> None:
-    """Set or update the auto-generated title for a session."""
-    with contextlib.suppress(Exception):
-        get_storage().update_session_title(session_id, title)
-
-
-# -- Workstream operations -----------------------------------------------------
+# -- Workstream management ----------------------------------------------------


 def register_workstream(
@@ -182,6 +79,95 @@ def list_workstreams(node_id: str | None = None, limit: int = 100) -> list[Any]:
        return []


+def list_workstreams_with_history(limit: int = 20) -> list[Any]:
+    """List workstreams that have conversation messages."""
+    try:
+        return get_storage().list_workstreams_with_history(limit)
+    except Exception:
+        return []
+
+
+def delete_workstream(ws_id: str) -> bool:
+    """Delete a workstream and all its conversations + config."""
+    try:
+        return get_storage().delete_workstream(ws_id)
+    except Exception:
+        return False
+
+
+def prune_workstreams(
+    retention_days: int = 90,
+    log_fn: Callable[[str], None] | None = None,
+) -> tuple[int, int]:
+    """Prune orphaned and stale workstreams."""
+    try:
+        orphans, stale = get_storage().prune_workstreams(retention_days)
+    except Exception:
+        return (0, 0)
+
+    if log_fn and (orphans or stale):
+        parts = []
+        if orphans:
+            parts.append(f"{orphans} empty workstream{'s' if orphans != 1 else ''}")
+        if stale:
+            parts.append(
+                f"{stale} workstream{'s' if stale != 1 else ''} older than {retention_days} days"
+            )
+        log_fn(f"[turnstone] Cleanup: removed {', '.join(parts)}.")
+
+    return (orphans, stale)
+
+
+def resolve_workstream(alias_or_id: str) -> str | None:
+    """Resolve an alias or ws_id (or prefix) to a full ws_id."""
+    try:
+        return get_storage().resolve_workstream(alias_or_id)
+    except Exception:
+        return None
+
+
+# -- Workstream config --------------------------------------------------------
+
+
+def save_workstream_config(ws_id: str, config: dict[str, str]) -> None:
+    """Persist workstream configuration key/value pairs."""
+    with contextlib.suppress(Exception):
+        get_storage().save_workstream_config(ws_id, config)
+
+
+def load_workstream_config(ws_id: str) -> dict[str, str]:
+    """Load workstream configuration."""
+    try:
+        return get_storage().load_workstream_config(ws_id)
+    except Exception:
+        return {}
+
+
+# -- Workstream metadata ------------------------------------------------------
+
+
+def set_workstream_alias(ws_id: str, alias: str) -> bool:
+    """Set a human-friendly alias. Returns False if alias is taken."""
+    try:
+        return get_storage().set_workstream_alias(ws_id, alias)
+    except Exception:
+        return False
+
+
+def get_workstream_display_name(ws_id: str) -> str | None:
+    """Return the alias (or title) for a workstream, or None if unset."""
+    try:
+        return get_storage().get_workstream_display_name(ws_id)
+    except Exception:
+        return None
+
+
+def update_workstream_title(ws_id: str, title: str) -> None:
+    """Set or update the auto-generated title for a workstream."""
+    with contextlib.suppress(Exception):
+        get_storage().update_workstream_title(ws_id, title)
+
+
 # -- Key-value store (memories) ------------------------------------------------


@@ -259,19 +259,13 @@ class MetricsCollector:

        # Per-workstream metrics (only when data is provided)
        if workstream_metrics:
-            # turnstone_workstream_info — exposes session_id as a label for joining,
-            # without propagating that high-cardinality label to counters.
-            lines.append(
-                "# HELP turnstone_workstream_info Workstream metadata"
-                " (join on session_id for per-session queries)"
-            )
+            lines.append("# HELP turnstone_workstream_info Workstream metadata")
            lines.append("# TYPE turnstone_workstream_info gauge")
            for wm in workstream_metrics:
                lstr = _fmt_labels(
                    {
                        "ws_id": wm["ws_id"],
                        "name": wm["name"],
-                        "session_id": wm["session_id"],
                    }
                )
                lines.append(f"turnstone_workstream_info{lstr} 1")
@@ -45,6 +45,14 @@ _OPENAI_CAPABILITIES: dict[str, ModelCapabilities] = {
        reasoning_effort_values=("minimal", "low", "medium", "high"),
        default_reasoning_effort="medium",
    ),
+    # GPT-5 pro — high reasoning only, extended output
+    "gpt-5-pro": ModelCapabilities(
+        context_window=400000,
+        max_output_tokens=272000,
+        supports_temperature=False,
+        reasoning_effort_values=("high",),
+        default_reasoning_effort="high",
+    ),
    # GPT-5.1 — temperature OK when reasoning_effort=none (default)
    "gpt-5.1": ModelCapabilities(
        context_window=400000,
@@ -59,6 +67,36 @@ _OPENAI_CAPABILITIES: dict[str, ModelCapabilities] = {
        reasoning_effort_values=("none", "low", "medium", "high", "xhigh"),
        default_reasoning_effort="none",
    ),
+    # GPT-5.2 pro — always-reasoning variant
+    "gpt-5.2-pro": ModelCapabilities(
+        context_window=400000,
+        max_output_tokens=128000,
+        supports_temperature=False,
+        reasoning_effort_values=("medium", "high", "xhigh"),
+        default_reasoning_effort="medium",
+    ),
+    # GPT-5.3 — same capabilities as 5.2 (matches gpt-5.3-chat-latest, codex)
+    "gpt-5.3": ModelCapabilities(
+        context_window=400000,
+        max_output_tokens=128000,
+        reasoning_effort_values=("none", "low", "medium", "high", "xhigh"),
+        default_reasoning_effort="none",
+    ),
+    # GPT-5.4 — 1M context window
+    "gpt-5.4": ModelCapabilities(
+        context_window=1050000,
+        max_output_tokens=128000,
+        reasoning_effort_values=("none", "low", "medium", "high", "xhigh"),
+        default_reasoning_effort="none",
+    ),
+    # GPT-5.4 pro — always-reasoning, 1M context
+    "gpt-5.4-pro": ModelCapabilities(
+        context_window=1050000,
+        max_output_tokens=128000,
+        supports_temperature=False,
+        reasoning_effort_values=("medium", "high", "xhigh"),
+        default_reasoning_effort="medium",
+    ),
    # O-series reasoning models
    "o1": ModelCapabilities(
        context_window=200000,
@@ -145,7 +183,11 @@ class OpenAIProvider:
            else:
                kwargs["temperature"] = temperature
        if caps.reasoning_effort_values and reasoning_effort and reasoning_effort != "none":
-            kwargs["reasoning_effort"] = reasoning_effort
+            # Validate against supported values; fall back to model default
+            if reasoning_effort in caps.reasoning_effort_values:
+                kwargs["reasoning_effort"] = reasoning_effort
+            elif caps.default_reasoning_effort and caps.default_reasoning_effort != "none":
+                kwargs["reasoning_effort"] = caps.default_reasoning_effort

    # -- web search ----------------------------------------------------------

@@ -26,29 +26,30 @@ import httpx

 from turnstone.core.config import get_tavily_key
 from turnstone.core.edit import find_occurrences, pick_nearest
+from turnstone.core.log import get_logger
 from turnstone.core.memory import (
    delete_memory,
-    delete_session,
-    get_session_name,
-    list_sessions,
+    delete_workstream,
+    get_workstream_display_name,
+    list_workstreams_with_history,
    load_memories,
-    load_session_config,
-    load_session_messages,
+    load_messages,
+    load_workstream_config,
    normalize_key,
-    register_session,
-    resolve_session,
+    resolve_workstream,
    save_memory,
    save_message,
-    save_session_config,
+    save_workstream_config,
    search_history,
    search_history_recent,
    search_memories,
-    set_session_alias,
-    update_session_title,
+    set_workstream_alias,
+    update_workstream_title,
 )
 from turnstone.core.providers import create_provider
 from turnstone.core.safety import is_command_blocked, sanitize_command
 from turnstone.core.sandbox import execute_math_sandboxed
+from turnstone.core.storage._registry import get_storage
 from turnstone.core.tools import (
    AGENT_AUTO_TOOLS,
    AGENT_TOOLS,
@@ -61,6 +62,8 @@ from turnstone.core.tools import (
 from turnstone.core.web import check_ssrf, strip_html
 from turnstone.ui.colors import DIM, GRAY, GREEN, RED, RESET, YELLOW, bold, cyan, dim

+log = get_logger(__name__)
+
 if TYPE_CHECKING:
    from collections.abc import Iterator

@@ -91,6 +94,43 @@ class SessionUI(Protocol):
    def on_rename(self, name: str) -> None: ...


+# ---------------------------------------------------------------------------
+# Notify auth helper (module-level, lazy-init)
+# ---------------------------------------------------------------------------
+
+_notify_token_manager: Any = None
+_notify_token_lock = threading.Lock()
+
+
+def _notify_auth_headers() -> dict[str, str]:
+    """Return Authorization headers for outbound notify requests."""
+    global _notify_token_manager
+
+    # Static token from env takes precedence
+    static_token = os.environ.get("TURNSTONE_CHANNEL_AUTH_TOKEN", "").strip()
+    if static_token:
+        return {"Authorization": f"Bearer {static_token}"}
+
+    # JWT via ServiceTokenManager
+    jwt_secret = os.environ.get("TURNSTONE_JWT_SECRET", "").strip()
+    if not jwt_secret:
+        return {}
+
+    with _notify_token_lock:
+        if _notify_token_manager is None:
+            from turnstone.core.auth import JWT_AUD_CHANNEL, ServiceTokenManager
+
+            _notify_token_manager = ServiceTokenManager(
+                user_id="system",
+                scopes=frozenset({"write"}),
+                source="service",
+                secret=jwt_secret,
+                audience=JWT_AUD_CHANNEL,
+            )
+    header: dict[str, str] = _notify_token_manager.bearer_header
+    return header
+
+
 # ---------------------------------------------------------------------------
 # ChatSession — the core engine
 # ---------------------------------------------------------------------------
@@ -151,10 +191,8 @@ class ChatSession:
        self.debug = False
        self.auto_approve = False
        self._node_id = node_id
-        self._ws_id = ws_id
-        self._session_id = uuid.uuid4().hex
+        self._ws_id = ws_id or uuid.uuid4().hex
        self._title_generated = False
-        register_session(self._session_id, node_id=self._node_id, ws_id=self._ws_id)
        self._read_files: set[str] = set()
        self.messages: list[dict[str, Any]] = []
        self._last_usage: dict[str, int] | None = None
@@ -162,6 +200,7 @@ class ChatSession:
        self._system_tokens = 0  # tokens for system_messages
        self._assistant_pending_tokens = 0
        self.creative_mode = False
+        self._notify_count = 0
        # MCP tool integration: merge external tools with built-in
        self._mcp_client = mcp_client
        if mcp_client:
@@ -177,17 +216,17 @@ class ChatSession:
        self._save_config()

    @property
-    def session_id(self) -> str:
-        return self._session_id
+    def ws_id(self) -> str:
+        return self._ws_id

    @property
    def model_alias(self) -> str | None:
        return self._model_alias

    def _save_config(self) -> None:
-        """Persist LLM-affecting config so resumed sessions behave identically."""
-        save_session_config(
-            self._session_id,
+        """Persist LLM-affecting config so resumed workstreams behave identically."""
+        save_workstream_config(
+            self._ws_id,
            {
                "temperature": str(self.temperature),
                "reasoning_effort": self.reasoning_effort,
@@ -256,32 +295,32 @@ class ChatSession:
            # Take first line, strip quotes
            title = raw.split("\n")[0].strip().strip('"').strip("'")
            if title:
-                update_session_title(self._session_id, title[:80])
+                update_workstream_title(self._ws_id, title[:80])
        except Exception:
            pass  # Title generation is non-critical

-    def resume_session(self, session_id: str) -> bool:
-        """Load messages from a previous session and resume it.
+    def resume(self, ws_id: str) -> bool:
+        """Load messages from a previous workstream and resume it.

        Replaces the current conversation with the loaded messages,
-        adopting the old session_id so new messages continue in the same
-        session.  Restores persisted config (temperature, reasoning_effort,
-        etc.) so the resumed session behaves identically to the original.
+        adopting the old ws_id so new messages continue in the same
+        workstream.  Restores persisted config (temperature, reasoning_effort,
+        etc.) so the resumed workstream behaves identically to the original.
        Returns True on success.
        """
-        messages = load_session_messages(session_id)
+        messages = load_messages(ws_id)
        if not messages:
            return False
-        self._session_id = session_id
+        self._ws_id = ws_id
        self.messages = messages
        self._read_files.clear()
        self._last_usage = None
-        self._title_generated = True  # don't re-title resumed sessions
+        self._title_generated = True  # don't re-title resumed workstreams
        self._msg_tokens = [
            max(1, int(self._msg_char_count(m) / self._chars_per_token)) for m in self.messages
        ]
        # Restore persisted config
-        config = load_session_config(session_id)
+        config = load_workstream_config(ws_id)
        if config:
            if "temperature" in config:
                self.temperature = float(config["temperature"])
@@ -468,9 +507,10 @@ class ChatSession:

    def send(self, user_input: str) -> None:
        """Send user input and handle the response loop (including tool calls)."""
+        self._notify_count = 0
        self.messages.append({"role": "user", "content": user_input})
        self._msg_tokens.append(max(1, int(len(user_input) / self._chars_per_token)))
-        save_message(self._session_id, "user", user_input)
+        save_message(self._ws_id, "user", user_input)

        try:
            while True:
@@ -506,9 +546,7 @@ class ChatSession:

                    provider_data = _json.dumps(assistant_msg["_provider_content"])
                if content or provider_data is not None:
-                    save_message(
-                        self._session_id, "assistant", content, provider_data=provider_data
-                    )
+                    save_message(self._ws_id, "assistant", content, provider_data=provider_data)
                if tc:
                    for call in tc:
                        fn = call.get("function", {})
@@ -519,7 +557,7 @@ class ChatSession:
                            "recall",
                        ):
                            save_message(
-                                self._session_id,
+                                self._ws_id,
                                "tool_call",
                                None,
                                name,
@@ -569,7 +607,7 @@ class ChatSession:
                        "recall",
                    ):
                        save_message(
-                            self._session_id,
+                            self._ws_id,
                            "tool_result",
                            output[:2000],
                            _tname,
@@ -1248,6 +1286,7 @@ class ChatSession:
            "remember": self._prepare_remember,
            "recall": self._prepare_recall,
            "forget": self._prepare_forget,
+            "notify": self._prepare_notify,
        }
        preparer = preparers.get(func_name)
        if not preparer:
@@ -2283,9 +2322,9 @@ class ChatSession:
    )

    def _exec_plan(self, item: dict[str, Any]) -> tuple[str, str]:
-        """Run a planning agent and write the result to .plan-<session_id>.md."""
+        """Run a planning agent and write the result to .plan-<ws_id>.md."""
        call_id, prompt = item["call_id"], item["prompt"]
-        plan_path = f".plan-{self._session_id}.md"
+        plan_path = f".plan-{self._ws_id}.md"

        # If plan was called before in this session, the previous assistant
        # tool_call + tool result are already in self.messages — pass them
@@ -2394,6 +2433,195 @@ class ChatSession:
        self.ui.on_tool_result(call_id, "recall", output)
        return call_id, output

+    # -- Notify tool -----------------------------------------------------------
+
+    def _prepare_notify(self, call_id: str, args: dict[str, Any]) -> dict[str, Any]:
+        """Prepare a channel notification."""
+        message = (args.get("message") or "").strip()
+        if not message:
+            return {
+                "call_id": call_id,
+                "func_name": "notify",
+                "header": "\u2717 notify: empty message",
+                "preview": "",
+                "needs_approval": False,
+                "error": "Error: message is required",
+            }
+        if len(message) > 2000:
+            return {
+                "call_id": call_id,
+                "func_name": "notify",
+                "header": "\u2717 notify: message too long",
+                "preview": "",
+                "needs_approval": False,
+                "error": "Error: message exceeds 2000 character limit",
+            }
+
+        username = (args.get("username") or "").strip()
+        channel_type = (args.get("channel_type") or "").strip()
+        channel_id = (args.get("channel_id") or "").strip()
+        title = (args.get("title") or "").strip()
+
+        has_username = bool(username)
+        has_direct = bool(channel_type and channel_id)
+
+        if has_username and has_direct:
+            return {
+                "call_id": call_id,
+                "func_name": "notify",
+                "header": "\u2717 notify: ambiguous target",
+                "preview": "",
+                "needs_approval": False,
+                "error": "Error: provide either username or channel_type+channel_id, not both",
+            }
+        if channel_type and not channel_id:
+            return {
+                "call_id": call_id,
+                "func_name": "notify",
+                "header": "\u2717 notify: incomplete target",
+                "preview": "",
+                "needs_approval": False,
+                "error": "Error: channel_id is required when channel_type is provided",
+            }
+        if channel_id and not channel_type:
+            return {
+                "call_id": call_id,
+                "func_name": "notify",
+                "header": "\u2717 notify: incomplete target",
+                "preview": "",
+                "needs_approval": False,
+                "error": "Error: channel_type is required when channel_id is provided",
+            }
+        if not has_username and not has_direct:
+            return {
+                "call_id": call_id,
+                "func_name": "notify",
+                "header": "\u2717 notify: no target",
+                "preview": "",
+                "needs_approval": False,
+                "error": "Error: provide username or channel_type+channel_id",
+            }
+
+        target_desc = f"@{username}" if has_username else f"{channel_type}:{channel_id}"
+
+        preview = message[:120] + ("..." if len(message) > 120 else "")
+        return {
+            "call_id": call_id,
+            "func_name": "notify",
+            "header": f"\u2709 notify \u2192 {target_desc}",
+            "preview": preview,
+            "needs_approval": False,
+            "execute": self._exec_notify,
+            "message": message,
+            "username": username,
+            "channel_type": channel_type,
+            "channel_id": channel_id,
+            "title": title,
+        }
+
+    _NOTIFY_MAX_RETRIES = 2
+    _NOTIFY_RETRY_DELAYS = (1.0, 3.0)
+
+    def _exec_notify(self, item: dict[str, Any]) -> tuple[str, str]:
+        """Send a notification directly to the channel gateway via HTTP."""
+        call_id = item["call_id"]
+
+        if self._notify_count >= 5:
+            msg = "Error: notification rate limit exceeded (max 5 per turn)"
+            self.ui.on_tool_result(call_id, "notify", msg)
+            return call_id, msg
+
+        target: dict[str, str] = {}
+        if item.get("username"):
+            target["username"] = item["username"]
+        else:
+            target["channel_type"] = item["channel_type"]
+            target["channel_id"] = item["channel_id"]
+
+        payload = {
+            "target": target,
+            "message": item["message"],
+            "title": item.get("title", ""),
+        }
+
+        # Build auth headers for service-to-service call
+        auth_headers = _notify_auth_headers()
+
+        # Retry loop: attempt delivery, re-query services on each retry
+        # in case a gateway comes back online between attempts.
+        for attempt in range(1 + self._NOTIFY_MAX_RETRIES):
+            storage = get_storage()
+            services = storage.list_services("channel", max_age_seconds=120)
+            if not services:
+                if attempt < self._NOTIFY_MAX_RETRIES:
+                    delay = self._NOTIFY_RETRY_DELAYS[attempt]
+                    log.warning(
+                        "notify.no_services",
+                        attempt=attempt + 1,
+                        max_retries=self._NOTIFY_MAX_RETRIES,
+                        retry_delay=delay,
+                    )
+                    time.sleep(delay)
+                    continue
+                log.warning("notify.no_services_exhausted")
+                msg = "Error: no channel gateway services available"
+                self.ui.on_tool_result(call_id, "notify", msg)
+                return call_id, msg
+
+            # Try first healthy gateway, fall back to next
+            last_error: str = ""
+            for svc in services:
+                url = svc["url"].rstrip("/") + "/v1/api/notify"
+                # SSRF guard: only allow http(s) URLs
+                if not url.startswith(("http://", "https://")):
+                    continue
+                try:
+                    resp = httpx.post(url, json=payload, timeout=10, headers=auth_headers)
+                    if resp.status_code < 300:
+                        # Check that at least one target was actually delivered
+                        try:
+                            data = resp.json()
+                        except Exception:
+                            last_error = "invalid gateway response"
+                            continue
+                        results = data.get("results") if isinstance(data, dict) else None
+                        if isinstance(results, list) and any(
+                            isinstance(r, dict) and r.get("status") == "sent" for r in results
+                        ):
+                            self._notify_count += 1
+                            msg = "Notification sent successfully"
+                            self.ui.on_tool_result(call_id, "notify", msg)
+                            return call_id, msg
+                        last_error = "no successful deliveries"
+                        continue
+                    last_error = f"HTTP {resp.status_code}"
+                except Exception as exc:
+                    last_error = type(exc).__name__
+                    continue  # try next gateway
+
+            # All gateways failed this attempt — retry if we have attempts left
+            if attempt < self._NOTIFY_MAX_RETRIES:
+                delay = self._NOTIFY_RETRY_DELAYS[attempt]
+                log.warning(
+                    "notify.all_gateways_failed",
+                    attempt=attempt + 1,
+                    max_retries=self._NOTIFY_MAX_RETRIES,
+                    last_error=last_error,
+                    gateway_count=len(services),
+                    retry_delay=delay,
+                )
+                time.sleep(delay)
+            else:
+                log.warning(
+                    "notify.delivery_failed",
+                    last_error=last_error,
+                    gateway_count=len(services),
+                )
+
+        msg = "Error: notification delivery failed"
+        self.ui.on_tool_result(call_id, "notify", msg)
+        return call_id, msg
+
    def _exec_write_file(self, item: dict[str, Any]) -> tuple[str, str]:
        """Write content to a file, creating parent directories as needed."""
        call_id = item["call_id"]
@@ -2675,29 +2903,31 @@ class ChatSession:
            self._read_files.clear()
            self._last_usage = None
            self._msg_tokens = []
-            self.ui.on_info("Context cleared (session preserved in database).")
+            self.ui.on_info("Context cleared (messages preserved in database).")

        elif cmd == "/new":
+            from turnstone.core.memory import register_workstream
+
            self.messages.clear()
            self._read_files.clear()
            self._last_usage = None
            self._msg_tokens = []
-            self._session_id = uuid.uuid4().hex
+            self._ws_id = uuid.uuid4().hex
            self._title_generated = False
-            register_session(self._session_id, node_id=self._node_id, ws_id=self._ws_id)
+            register_workstream(self._ws_id, node_id=self._node_id)
            self._save_config()
-            self.ui.on_info("New session started.")
+            self.ui.on_info("New workstream started.")

-        elif cmd == "/sessions":
-            rows = list_sessions(limit=20)
+        elif cmd == "/workstreams":
+            rows = list_workstreams_with_history(limit=20)
            if not rows:
-                self.ui.on_info("No saved sessions.")
+                self.ui.on_info("No saved workstreams.")
            else:
-                lines = ["Sessions:\n"]
-                for sid, alias, title, _created, updated, count, *_extra in rows:
-                    display_name = alias or sid
+                lines = ["Workstreams:\n"]
+                for wid, alias, title, _created, updated, count, *_extra in rows:
+                    display_name = alias or wid
                    display_title = f"  {title}" if title else ""
-                    marker = " *" if sid == self._session_id else "  "
+                    marker = " *" if wid == self._ws_id else "  "
                    lines.append(
                        f" {marker} {bold(display_name)}{display_title}  "
                        f"{dim(f'{count} msgs, {updated}')}"
@@ -2707,30 +2937,29 @@ class ChatSession:
        elif cmd == "/resume":
            if not arg:
                self.ui.on_info(
-                    "Usage: /resume <alias_or_session_id>\n"
-                    "Use /sessions to list available sessions."
+                    "Usage: /resume <alias_or_ws_id>\nUse /workstreams to list available workstreams."
                )
            else:
-                target_id = resolve_session(arg.strip())
+                target_id = resolve_workstream(arg.strip())
                if not target_id:
-                    self.ui.on_info(f"Session not found: {arg.strip()}")
-                elif target_id == self._session_id:
-                    self.ui.on_info("Already in that session.")
-                elif self.resume_session(target_id):
+                    self.ui.on_info(f"Workstream not found: {arg.strip()}")
+                elif target_id == self._ws_id:
+                    self.ui.on_info("Already in that workstream.")
+                elif self.resume(target_id):
                    self.ui.on_info(
-                        f"Resumed session {bold(target_id)} ({len(self.messages)} messages loaded)"
+                        f"Resumed {bold(target_id)} ({len(self.messages)} messages loaded)"
                    )
-                    name = get_session_name(target_id)
+                    name = get_workstream_display_name(target_id)
                    if name:
                        self.ui.on_rename(name)
                else:
-                    self.ui.on_info(f"Session {arg.strip()} has no messages.")
+                    self.ui.on_info(f"Workstream {arg.strip()} has no messages.")

        elif cmd == "/name":
            if not arg:
-                self.ui.on_info(f"Current session: {self._session_id}")
-            elif set_session_alias(self._session_id, arg.strip()):
-                self.ui.on_info(f"Session named: {bold(arg.strip())}")
+                self.ui.on_info(f"Current workstream: {self._ws_id}")
+            elif set_workstream_alias(self._ws_id, arg.strip()):
+                self.ui.on_info(f"Workstream named: {bold(arg.strip())}")
                self.ui.on_rename(arg.strip())
            else:
                self.ui.on_info(f"Alias '{arg.strip()}' is already in use.")
@@ -2738,18 +2967,18 @@ class ChatSession:
        elif cmd == "/delete":
            if not arg:
                self.ui.on_info(
-                    "Usage: /delete <alias_or_session_id>\nUse /sessions to list sessions."
+                    "Usage: /delete <alias_or_ws_id>\nUse /workstreams to list workstreams."
                )
            else:
-                target_id = resolve_session(arg.strip())
+                target_id = resolve_workstream(arg.strip())
                if not target_id:
-                    self.ui.on_info(f"Session not found: {arg.strip()}")
-                elif target_id == self._session_id:
-                    self.ui.on_info("Cannot delete the active session.")
-                elif delete_session(target_id):
-                    self.ui.on_info(f"Deleted session {arg.strip()}")
+                    self.ui.on_info(f"Workstream not found: {arg.strip()}")
+                elif target_id == self._ws_id:
+                    self.ui.on_info("Cannot delete the active workstream.")
+                elif delete_workstream(target_id):
+                    self.ui.on_info(f"Deleted workstream {arg.strip()}")
                else:
-                    self.ui.on_info(f"Failed to delete session {arg.strip()}")
+                    self.ui.on_info(f"Failed to delete workstream {arg.strip()}")

        elif cmd == "/history":
            query = arg.strip() if arg else None
@@ -2877,13 +3106,13 @@ class ChatSession:
                    [
                        "── Slash Commands ─────────────────────────────────────",
                        "  /instructions <text>   Set developer instructions",
-                        "  /clear                 Clear context (session preserved in database)",
-                        "  /new                   Start a new session (old session stays resumable)",
+                        "  /clear                 Clear context (workstream preserved in database)",
+                        "  /new                   Start a new workstream (old one stays resumable)",
                        "",
-                        "  /sessions              List saved sessions",
-                        "  /resume <id|alias>     Resume a previous session",
-                        "  /name <alias>          Name the current session",
-                        "  /delete <id|alias>     Delete a saved session",
+                        "  /workstreams           List saved workstreams",
+                        "  /resume <id|alias>     Resume a previous workstream",
+                        "  /name <alias>          Name the current workstream",
+                        "  /delete <id|alias>     Delete a saved workstream",
                        "",
                        "  /history [query]       Search conversation history (or show recent)",
                        "  /compact               Compact conversation (summarize old messages)",
@@ -13,9 +13,8 @@ from turnstone.core.storage._schema import (
    conversations,
    memories,
    metadata,
-    session_config,
-    sessions,
    users,
+    workstream_config,
    workstreams,
 )
 from turnstone.core.storage._sqlite import _reconstruct_messages
@@ -38,40 +37,11 @@ class PostgreSQLBackend:
        if create_tables:
            metadata.create_all(self._engine)

-    # -- Core session operations -----------------------------------------------
-
-    def register_session(
-        self,
-        session_id: str,
-        title: str | None = None,
-        node_id: str | None = None,
-        ws_id: str | None = None,
-        user_id: str | None = None,
-    ) -> None:
-        now = datetime.now(UTC).strftime("%Y-%m-%dT%H:%M:%S")
-        with self._engine.connect() as conn:
-            # Use dialect-neutral upsert pattern
-            existing = conn.execute(
-                sa.select(sessions.c.session_id).where(sessions.c.session_id == session_id)
-            ).fetchone()
-            if not existing:
-                conn.execute(
-                    sa.insert(sessions),
-                    {
-                        "session_id": session_id,
-                        "title": title,
-                        "node_id": node_id,
-                        "ws_id": ws_id,
-                        "user_id": user_id,
-                        "created": now,
-                        "updated": now,
-                    },
-                )
-            conn.commit()
+    # -- Core conversation operations ------------------------------------------

    def save_message(
        self,
-        session_id: str,
+        ws_id: str,
        role: str,
        content: str | None,
        tool_name: str | None = None,
@@ -84,7 +54,7 @@ class PostgreSQLBackend:
            conn.execute(
                sa.insert(conversations),
                {
-                    "session_id": session_id,
+                    "ws_id": ws_id,
                    "timestamp": now,
                    "role": role,
                    "content": content,
@@ -95,11 +65,11 @@ class PostgreSQLBackend:
                },
            )
            conn.execute(
-                sa.update(sessions).where(sessions.c.session_id == session_id).values(updated=now)
+                sa.update(workstreams).where(workstreams.c.ws_id == ws_id).values(updated=now)
            )
            conn.commit()

-    def load_session_messages(self, session_id: str) -> list[dict[str, Any]]:
+    def load_messages(self, ws_id: str) -> list[dict[str, Any]]:
        with self._engine.connect() as conn:
            rows = conn.execute(
                sa.select(
@@ -110,174 +80,149 @@ class PostgreSQLBackend:
                    conversations.c.tool_call_id,
                    conversations.c.provider_data,
                )
-                .where(conversations.c.session_id == session_id)
+                .where(conversations.c.ws_id == ws_id)
                .order_by(conversations.c.id)
            ).fetchall()
-        return _reconstruct_messages(list(rows), session_id)
+        return _reconstruct_messages(list(rows), ws_id)

-    # -- Session management ----------------------------------------------------
+    # -- Workstream management -------------------------------------------------

-    def list_sessions(self, limit: int = 20) -> list[Any]:
+    def list_workstreams_with_history(self, limit: int = 20) -> list[Any]:
        with self._engine.connect() as conn:
            return list(
                conn.execute(
                    sa.text(
-                        "SELECT s.session_id, s.alias, s.title, s.created, s.updated, "
+                        "SELECT w.ws_id, w.alias, w.title, w.created, w.updated, "
                        "(SELECT COUNT(*) FROM conversations c "
-                        " WHERE c.session_id = s.session_id), "
-                        "s.node_id, s.ws_id "
-                        "FROM sessions s "
+                        " WHERE c.ws_id = w.ws_id), "
+                        "w.node_id "
+                        "FROM workstreams w "
                        "WHERE EXISTS "
-                        "  (SELECT 1 FROM conversations c WHERE c.session_id = s.session_id) "
-                        "ORDER BY s.updated DESC LIMIT :limit"
+                        "  (SELECT 1 FROM conversations c WHERE c.ws_id = w.ws_id) "
+                        "ORDER BY w.updated DESC LIMIT :limit"
                    ),
                    {"limit": limit},
                ).fetchall()
            )

-    def delete_session(self, session_id: str) -> bool:
-        with self._engine.connect() as conn:
-            conn.execute(sa.delete(conversations).where(conversations.c.session_id == session_id))
-            conn.execute(sa.delete(session_config).where(session_config.c.session_id == session_id))
-            conn.execute(sa.delete(sessions).where(sessions.c.session_id == session_id))
-            conn.commit()
-            return True
-
-    def prune_sessions(self, retention_days: int = 90) -> tuple[int, int]:
+    def prune_workstreams(self, retention_days: int = 90) -> tuple[int, int]:
        orphans = stale = 0
        with self._engine.connect() as conn:
-            # 1. Remove sessions with no messages
+            # 1. Remove workstreams with no messages
            orphan_rows = conn.execute(
                sa.text(
-                    "SELECT session_id FROM sessions "
+                    "SELECT ws_id FROM workstreams "
                    "WHERE NOT EXISTS "
                    "  (SELECT 1 FROM conversations c "
-                    "   WHERE c.session_id = sessions.session_id)"
+                    "   WHERE c.ws_id = workstreams.ws_id)"
                )
            ).fetchall()
            orphan_ids = [r[0] for r in orphan_rows]
            if orphan_ids:
                conn.execute(
-                    sa.delete(session_config).where(session_config.c.session_id.in_(orphan_ids))
+                    sa.delete(workstream_config).where(workstream_config.c.ws_id.in_(orphan_ids))
                )
                result = conn.execute(
-                    sa.delete(sessions).where(sessions.c.session_id.in_(orphan_ids))
+                    sa.delete(workstreams).where(workstreams.c.ws_id.in_(orphan_ids))
                )
                orphans = result.rowcount

-            # 2. Remove old unnamed sessions
+            # 2. Remove old unnamed workstreams
            if retention_days > 0:
                cutoff = (datetime.now(UTC) - timedelta(days=retention_days)).strftime(
                    "%Y-%m-%dT%H:%M:%S"
                )
                stale_rows = conn.execute(
-                    sa.select(sessions.c.session_id).where(
-                        sessions.c.alias.is_(None),
-                        sessions.c.updated < cutoff,
+                    sa.select(workstreams.c.ws_id).where(
+                        workstreams.c.alias.is_(None),
+                        workstreams.c.updated < cutoff,
                    )
                ).fetchall()
                stale_ids = [r[0] for r in stale_rows]
                if stale_ids:
                    conn.execute(
-                        sa.delete(session_config).where(session_config.c.session_id.in_(stale_ids))
+                        sa.delete(conversations).where(conversations.c.ws_id.in_(stale_ids))
+                    )
+                    conn.execute(
+                        sa.delete(workstream_config).where(workstream_config.c.ws_id.in_(stale_ids))
                    )
                    result = conn.execute(
-                        sa.delete(sessions).where(sessions.c.session_id.in_(stale_ids))
+                        sa.delete(workstreams).where(workstreams.c.ws_id.in_(stale_ids))
                    )
                    stale = result.rowcount

            conn.commit()
        return (orphans, stale)

-    def resolve_session(self, alias_or_id: str) -> str | None:
+    def resolve_workstream(self, alias_or_id: str) -> str | None:
        with self._engine.connect() as conn:
            # 1. Exact alias
            row = conn.execute(
-                sa.select(sessions.c.session_id).where(sessions.c.alias == alias_or_id)
+                sa.select(workstreams.c.ws_id).where(workstreams.c.alias == alias_or_id)
            ).fetchone()
            if row:
                return str(row[0])
-            # 2. Exact session_id
+            # 2. Exact ws_id
            row = conn.execute(
-                sa.select(sessions.c.session_id).where(sessions.c.session_id == alias_or_id)
+                sa.select(workstreams.c.ws_id).where(workstreams.c.ws_id == alias_or_id)
            ).fetchone()
            if row:
                return str(row[0])
            # 3. Prefix match
            rows = conn.execute(
-                sa.select(sessions.c.session_id).where(
-                    sessions.c.session_id.like(alias_or_id + "%")
-                )
+                sa.select(workstreams.c.ws_id).where(workstreams.c.ws_id.like(alias_or_id + "%"))
            ).fetchall()
            if len(rows) == 1:
                return str(rows[0][0])
-            # 4. Legacy: check conversations
-            row = conn.execute(
-                sa.select(sa.distinct(conversations.c.session_id))
-                .where(conversations.c.session_id == alias_or_id)
-                .limit(1)
-            ).fetchone()
-            if row:
-                now = datetime.now(UTC).strftime("%Y-%m-%dT%H:%M:%S")
-                existing = conn.execute(
-                    sa.select(sessions.c.session_id).where(sessions.c.session_id == row[0])
-                ).fetchone()
-                if not existing:
-                    conn.execute(
-                        sa.insert(sessions),
-                        {"session_id": row[0], "created": now, "updated": now},
-                    )
-                    conn.commit()
-                return str(row[0])
            return None

-    # -- Session config --------------------------------------------------------
+    # -- Workstream config -----------------------------------------------------

-    def save_session_config(self, session_id: str, config: dict[str, str]) -> None:
+    def save_workstream_config(self, ws_id: str, config: dict[str, str]) -> None:
        with self._engine.connect() as conn:
            for key, value in config.items():
                # Upsert: delete + insert
                conn.execute(
-                    sa.delete(session_config).where(
-                        session_config.c.session_id == session_id,
-                        session_config.c.key == key,
+                    sa.delete(workstream_config).where(
+                        workstream_config.c.ws_id == ws_id,
+                        workstream_config.c.key == key,
                    )
                )
                conn.execute(
-                    sa.insert(session_config),
-                    {"session_id": session_id, "key": key, "value": value},
+                    sa.insert(workstream_config),
+                    {"ws_id": ws_id, "key": key, "value": value},
                )
            conn.commit()

-    def load_session_config(self, session_id: str) -> dict[str, str]:
+    def load_workstream_config(self, ws_id: str) -> dict[str, str]:
        with self._engine.connect() as conn:
            rows = conn.execute(
-                sa.select(session_config.c.key, session_config.c.value).where(
-                    session_config.c.session_id == session_id
+                sa.select(workstream_config.c.key, workstream_config.c.value).where(
+                    workstream_config.c.ws_id == ws_id
                )
            ).fetchall()
            return {row[0]: row[1] for row in rows}

-    # -- Session metadata ------------------------------------------------------
+    # -- Workstream metadata ---------------------------------------------------

-    def set_session_alias(self, session_id: str, alias: str) -> bool:
+    def set_workstream_alias(self, ws_id: str, alias: str) -> bool:
        with self._engine.connect() as conn:
            existing = conn.execute(
-                sa.select(sessions.c.session_id).where(sessions.c.alias == alias)
+                sa.select(workstreams.c.ws_id).where(workstreams.c.alias == alias)
            ).fetchone()
-            if existing and existing[0] != session_id:
+            if existing and existing[0] != ws_id:
                return False
            conn.execute(
-                sa.update(sessions).where(sessions.c.session_id == session_id).values(alias=alias)
+                sa.update(workstreams).where(workstreams.c.ws_id == ws_id).values(alias=alias)
            )
            conn.commit()
            return True

-    def get_session_name(self, session_id: str) -> str | None:
+    def get_workstream_display_name(self, ws_id: str) -> str | None:
        with self._engine.connect() as conn:
            row = conn.execute(
-                sa.select(sessions.c.alias, sessions.c.title).where(
-                    sessions.c.session_id == session_id
+                sa.select(workstreams.c.alias, workstreams.c.title).where(
+                    workstreams.c.ws_id == ws_id
                )
            ).fetchone()
            if row:
@@ -285,10 +230,10 @@ class PostgreSQLBackend:
                return str(value) if value is not None else None
        return None

-    def update_session_title(self, session_id: str, title: str) -> None:
+    def update_workstream_title(self, ws_id: str, title: str) -> None:
        with self._engine.connect() as conn:
            conn.execute(
-                sa.update(sessions).where(sessions.c.session_id == session_id).values(title=title)
+                sa.update(workstreams).where(workstreams.c.ws_id == ws_id).values(title=title)
            )
            conn.commit()

@@ -359,6 +304,8 @@ class PostgreSQLBackend:
        name: str = "",
        state: str = "idle",
        user_id: str | None = None,
+        alias: str | None = None,
+        title: str | None = None,
    ) -> None:
        now = datetime.now(UTC).strftime("%Y-%m-%dT%H:%M:%S")
        with self._engine.connect() as conn:
@@ -374,6 +321,8 @@ class PostgreSQLBackend:
                        "user_id": user_id,
                        "name": name,
                        "state": state,
+                        "alias": alias,
+                        "title": title,
                        "created": now,
                        "updated": now,
                    },
@@ -402,6 +351,8 @@ class PostgreSQLBackend:

    def delete_workstream(self, ws_id: str) -> bool:
        with self._engine.connect() as conn:
+            conn.execute(sa.delete(conversations).where(conversations.c.ws_id == ws_id))
+            conn.execute(sa.delete(workstream_config).where(workstream_config.c.ws_id == ws_id))
            result = conn.execute(sa.delete(workstreams).where(workstreams.c.ws_id == ws_id))
            conn.commit()
            return result.rowcount > 0
@@ -436,7 +387,7 @@ class PostgreSQLBackend:
                return list(
                    conn.execute(
                        sa.text(
-                            "SELECT c.timestamp, c.session_id, c.role, c.content, c.tool_name "
+                            "SELECT c.timestamp, c.ws_id, c.role, c.content, c.tool_name "
                            "FROM conversations c "
                            "WHERE to_tsvector('english', COALESCE(c.content, '')) "
                            "   @@ plainto_tsquery('english', :query) "
@@ -452,7 +403,7 @@ class PostgreSQLBackend:
                return list(
                    conn.execute(
                        sa.text(
-                            "SELECT timestamp, session_id, role, content, tool_name "
+                            "SELECT timestamp, ws_id, role, content, tool_name "
                            "FROM conversations WHERE content ILIKE :pattern "
                            "ORDER BY timestamp DESC LIMIT :limit"
                        ),
@@ -466,22 +417,13 @@ class PostgreSQLBackend:
            return list(
                conn.execute(
                    sa.text(
-                        "SELECT timestamp, session_id, role, content, tool_name "
+                        "SELECT timestamp, ws_id, role, content, tool_name "
                        "FROM conversations ORDER BY timestamp DESC LIMIT :limit"
                    ),
                    {"limit": capped},
                ).fetchall()
            )

-    # -- Session lookup by workstream ------------------------------------------
-
-    def get_session_id_by_ws(self, ws_id: str) -> str | None:
-        with self._engine.connect() as conn:
-            row = conn.execute(
-                sa.select(sessions.c.session_id).where(sessions.c.ws_id == ws_id)
-            ).fetchone()
-            return str(row[0]) if row else None
-
    # -- User identity operations -----------------------------------------------

    def create_user(
@@ -879,6 +821,268 @@ class PostgreSQLBackend:
            conn.commit()
            return result.rowcount > 0

+    # -- Scheduled tasks -------------------------------------------------------
+
+    def create_scheduled_task(
+        self,
+        task_id: str,
+        name: str,
+        description: str,
+        schedule_type: str,
+        cron_expr: str,
+        at_time: str,
+        target_mode: str,
+        model: str,
+        initial_message: str,
+        auto_approve: bool,
+        auto_approve_tools: list[str],
+        created_by: str,
+        next_run: str,
+    ) -> None:
+        from sqlalchemy.dialects import postgresql
+
+        from turnstone.core.storage._schema import scheduled_tasks
+
+        now = datetime.now(UTC).strftime("%Y-%m-%dT%H:%M:%S")
+        with self._engine.connect() as conn:
+            conn.execute(
+                postgresql.insert(scheduled_tasks)
+                .values(
+                    task_id=task_id,
+                    name=name,
+                    description=description,
+                    schedule_type=schedule_type,
+                    cron_expr=cron_expr,
+                    at_time=at_time,
+                    target_mode=target_mode,
+                    model=model,
+                    initial_message=initial_message,
+                    auto_approve=1 if auto_approve else 0,
+                    auto_approve_tools=",".join(auto_approve_tools),
+                    enabled=1,
+                    created_by=created_by,
+                    next_run=next_run,
+                    created=now,
+                    updated=now,
+                )
+                .on_conflict_do_nothing()
+            )
+            conn.commit()
+
+    def get_scheduled_task(self, task_id: str) -> dict[str, Any] | None:
+        from turnstone.core.storage._schema import scheduled_tasks
+
+        with self._engine.connect() as conn:
+            row = conn.execute(
+                sa.select(scheduled_tasks).where(scheduled_tasks.c.task_id == task_id)
+            ).fetchone()
+            if row is None:
+                return None
+            return dict(row._mapping)
+
+    def list_scheduled_tasks(self) -> list[dict[str, Any]]:
+        from turnstone.core.storage._schema import scheduled_tasks
+
+        with self._engine.connect() as conn:
+            rows = conn.execute(
+                sa.select(scheduled_tasks).order_by(scheduled_tasks.c.created.desc())
+            ).fetchall()
+            return [dict(r._mapping) for r in rows]
+
+    _UPDATABLE_TASK_FIELDS = frozenset(
+        {
+            "name",
+            "description",
+            "schedule_type",
+            "cron_expr",
+            "at_time",
+            "target_mode",
+            "model",
+            "initial_message",
+            "auto_approve",
+            "auto_approve_tools",
+            "enabled",
+            "last_run",
+            "next_run",
+            "updated",
+        }
+    )
+
+    def update_scheduled_task(self, task_id: str, **fields: Any) -> bool:
+        from turnstone.core.storage._schema import scheduled_tasks
+
+        fields = {k: v for k, v in fields.items() if k in self._UPDATABLE_TASK_FIELDS}
+        fields["updated"] = datetime.now(UTC).strftime("%Y-%m-%dT%H:%M:%S")
+        if "auto_approve" in fields:
+            fields["auto_approve"] = 1 if fields["auto_approve"] else 0
+        if "auto_approve_tools" in fields and isinstance(fields["auto_approve_tools"], list):
+            fields["auto_approve_tools"] = ",".join(fields["auto_approve_tools"])
+        if "enabled" in fields:
+            fields["enabled"] = 1 if fields["enabled"] else 0
+        with self._engine.connect() as conn:
+            result = conn.execute(
+                sa.update(scheduled_tasks)
+                .where(scheduled_tasks.c.task_id == task_id)
+                .values(**fields)
+            )
+            conn.commit()
+            return result.rowcount > 0
+
+    def delete_scheduled_task(self, task_id: str) -> bool:
+        from turnstone.core.storage._schema import scheduled_task_runs, scheduled_tasks
+
+        with self._engine.connect() as conn:
+            conn.execute(
+                sa.delete(scheduled_task_runs).where(scheduled_task_runs.c.task_id == task_id)
+            )
+            result = conn.execute(
+                sa.delete(scheduled_tasks).where(scheduled_tasks.c.task_id == task_id)
+            )
+            conn.commit()
+            return result.rowcount > 0
+
+    def list_due_tasks(self, now: str) -> list[dict[str, Any]]:
+        from turnstone.core.storage._schema import scheduled_tasks
+
+        with self._engine.connect() as conn:
+            rows = conn.execute(
+                sa.select(scheduled_tasks)
+                .where(
+                    (scheduled_tasks.c.enabled == 1)
+                    & (scheduled_tasks.c.next_run <= now)
+                    & (scheduled_tasks.c.next_run != "")
+                )
+                .order_by(scheduled_tasks.c.next_run)
+                .limit(100)
+            ).fetchall()
+            return [dict(r._mapping) for r in rows]
+
+    def record_task_run(
+        self,
+        run_id: str,
+        task_id: str,
+        node_id: str,
+        ws_id: str,
+        correlation_id: str,
+        started: str,
+        status: str,
+        error: str,
+    ) -> None:
+        from turnstone.core.storage._schema import scheduled_task_runs
+
+        with self._engine.connect() as conn:
+            conn.execute(
+                sa.insert(scheduled_task_runs),
+                {
+                    "run_id": run_id,
+                    "task_id": task_id,
+                    "node_id": node_id,
+                    "ws_id": ws_id,
+                    "correlation_id": correlation_id,
+                    "started": started,
+                    "status": status,
+                    "error": error,
+                },
+            )
+            conn.commit()
+
+    def list_task_runs(self, task_id: str, limit: int = 50) -> list[dict[str, Any]]:
+        from turnstone.core.storage._schema import scheduled_task_runs
+
+        with self._engine.connect() as conn:
+            rows = conn.execute(
+                sa.select(scheduled_task_runs)
+                .where(scheduled_task_runs.c.task_id == task_id)
+                .order_by(scheduled_task_runs.c.started.desc())
+                .limit(limit)
+            ).fetchall()
+            return [dict(r._mapping) for r in rows]
+
+    def prune_task_runs(self, retention_days: int = 90) -> int:
+        from datetime import timedelta
+
+        from turnstone.core.storage._schema import scheduled_task_runs
+
+        cutoff = (datetime.now(UTC) - timedelta(days=retention_days)).strftime("%Y-%m-%dT%H:%M:%S")
+        with self._engine.connect() as conn:
+            result = conn.execute(
+                sa.delete(scheduled_task_runs).where(scheduled_task_runs.c.started < cutoff)
+            )
+            conn.commit()
+            return result.rowcount
+
+    # -- Service registry ------------------------------------------------------
+
+    def register_service(
+        self, service_type: str, service_id: str, url: str, metadata: str = "{}"
+    ) -> None:
+        from turnstone.core.storage._schema import services
+
+        now = datetime.now(UTC).strftime("%Y-%m-%dT%H:%M:%S")
+        with self._engine.connect() as conn:
+            from sqlalchemy.dialects.postgresql import insert as pg_insert
+
+            stmt = pg_insert(services).values(
+                service_type=service_type,
+                service_id=service_id,
+                url=url,
+                metadata=metadata,
+                last_heartbeat=now,
+                created=now,
+            )
+            stmt = stmt.on_conflict_do_update(
+                index_elements=[services.c.service_type, services.c.service_id],
+                set_={"url": url, "metadata": metadata, "last_heartbeat": now},
+            )
+            conn.execute(stmt)
+            conn.commit()
+
+    def heartbeat_service(self, service_type: str, service_id: str) -> bool:
+        from turnstone.core.storage._schema import services
+
+        now = datetime.now(UTC).strftime("%Y-%m-%dT%H:%M:%S")
+        with self._engine.connect() as conn:
+            result = conn.execute(
+                sa.update(services)
+                .where(
+                    (services.c.service_type == service_type)
+                    & (services.c.service_id == service_id)
+                )
+                .values(last_heartbeat=now)
+            )
+            conn.commit()
+            return result.rowcount > 0
+
+    def list_services(self, service_type: str, max_age_seconds: int = 120) -> list[dict[str, str]]:
+        from turnstone.core.storage._schema import services
+
+        cutoff = (datetime.now(UTC) - timedelta(seconds=max_age_seconds)).strftime(
+            "%Y-%m-%dT%H:%M:%S"
+        )
+        with self._engine.connect() as conn:
+            rows = conn.execute(
+                sa.select(services)
+                .where(
+                    (services.c.service_type == service_type)
+                    & (services.c.last_heartbeat >= cutoff)
+                )
+                .order_by(services.c.last_heartbeat.desc())
+            ).fetchall()
+            return [dict(r._mapping) for r in rows]
+
+    def deregister_service(self, service_type: str, service_id: str) -> bool:
+        from turnstone.core.storage._schema import services
+
+        with self._engine.connect() as conn:
+            result = conn.execute(
+                sa.delete(services).where(
+                    (services.c.service_type == service_type)
+                    & (services.c.service_id == service_id)
+                )
+            )
+            conn.commit()
+            return result.rowcount > 0
+
    # -- Lifecycle -------------------------------------------------------------

    def close(self) -> None:
@@ -9,26 +9,15 @@ from typing import Any, Protocol, runtime_checkable
 class StorageBackend(Protocol):
    """Protocol that every storage backend adapter must implement.

-    Provides session management, conversation persistence, key-value storage
+    Provides workstream management, conversation persistence, key-value storage
    (for memories), and full-text search.
    """

-    # -- Core session operations -----------------------------------------------
-
-    def register_session(
-        self,
-        session_id: str,
-        title: str | None = None,
-        node_id: str | None = None,
-        ws_id: str | None = None,
-        user_id: str | None = None,
-    ) -> None:
-        """Create a sessions row for a new session (no-op if already exists)."""
-        ...
+    # -- Core conversation operations ------------------------------------------

    def save_message(
        self,
-        session_id: str,
+        ws_id: str,
        role: str,
        content: str | None,
        tool_name: str | None = None,
@@ -39,50 +28,46 @@ class StorageBackend(Protocol):
        """Log a message to the conversations table."""
        ...

-    def load_session_messages(self, session_id: str) -> list[dict[str, Any]]:
-        """Load messages for a session and reconstruct OpenAI message format."""
+    def load_messages(self, ws_id: str) -> list[dict[str, Any]]:
+        """Load messages for a workstream and reconstruct OpenAI message format."""
        ...

-    # -- Session management ----------------------------------------------------
+    # -- Workstream management -------------------------------------------------

-    def list_sessions(self, limit: int = 20) -> list[Any]:
-        """List recent sessions with message counts, ordered by updated DESC."""
+    def list_workstreams_with_history(self, limit: int = 20) -> list[Any]:
+        """List workstreams that have messages, ordered by updated DESC."""
        ...

-    def delete_session(self, session_id: str) -> bool:
-        """Delete a session and all its messages. Returns True on success."""
+    def prune_workstreams(self, retention_days: int = 90) -> tuple[int, int]:
+        """Remove orphaned + stale unnamed workstreams. Returns (orphans, stale)."""
        ...

-    def prune_sessions(self, retention_days: int = 90) -> tuple[int, int]:
-        """Remove orphaned + stale unnamed sessions. Returns (orphans, stale)."""
+    def resolve_workstream(self, alias_or_id: str) -> str | None:
+        """Resolve an alias or ws_id (or prefix) to a full ws_id."""
        ...

-    def resolve_session(self, alias_or_id: str) -> str | None:
-        """Resolve an alias or session_id (or prefix) to a full session_id."""
+    # -- Workstream config -----------------------------------------------------
+
+    def save_workstream_config(self, ws_id: str, config: dict[str, str]) -> None:
+        """Persist workstream configuration key/value pairs."""
        ...

-    # -- Session config --------------------------------------------------------
-
-    def save_session_config(self, session_id: str, config: dict[str, str]) -> None:
-        """Persist session configuration key/value pairs."""
+    def load_workstream_config(self, ws_id: str) -> dict[str, str]:
+        """Load workstream configuration. Returns empty dict if none stored."""
        ...

-    def load_session_config(self, session_id: str) -> dict[str, str]:
-        """Load session configuration. Returns empty dict if none stored."""
-        ...
+    # -- Workstream metadata ---------------------------------------------------

-    # -- Session metadata ------------------------------------------------------
-
-    def set_session_alias(self, session_id: str, alias: str) -> bool:
+    def set_workstream_alias(self, ws_id: str, alias: str) -> bool:
        """Set a human-friendly alias. Returns False if alias is taken."""
        ...

-    def get_session_name(self, session_id: str) -> str | None:
-        """Return the alias (or title) for a session, or None if unset."""
+    def get_workstream_display_name(self, ws_id: str) -> str | None:
+        """Return the alias (or title) for a workstream, or None if unset."""
        ...

-    def update_session_title(self, session_id: str, title: str) -> None:
-        """Set or update the auto-generated title for a session."""
+    def update_workstream_title(self, ws_id: str, title: str) -> None:
+        """Set or update the auto-generated title for a workstream."""
        ...

    # -- Generic key-value store (backs memories table) ------------------------
@@ -116,6 +101,8 @@ class StorageBackend(Protocol):
        name: str = "",
        state: str = "idle",
        user_id: str | None = None,
+        alias: str | None = None,
+        title: str | None = None,
    ) -> None:
        """Create a workstreams row (no-op if already exists)."""
        ...
@@ -129,7 +116,7 @@ class StorageBackend(Protocol):
        ...

    def delete_workstream(self, ws_id: str) -> bool:
-        """Delete a workstream. Returns True on success."""
+        """Delete a workstream and all its conversations + config."""
        ...

    def list_workstreams(self, node_id: str | None = None, limit: int = 100) -> list[Any]:
@@ -139,7 +126,7 @@ class StorageBackend(Protocol):
    # -- Conversation search ---------------------------------------------------

    def search_history(self, query: str, limit: int = 20) -> list[Any]:
-        """Search conversation history. Returns (timestamp, session_id, role, content, tool_name)."""
+        """Search conversation history. Returns (timestamp, ws_id, role, content, tool_name)."""
        ...

    def search_history_recent(self, limit: int = 20) -> list[Any]:
@@ -219,12 +206,6 @@ class StorageBackend(Protocol):
        """Remove a channel user mapping. Returns True if existed."""
        ...

-    # -- Session lookup by workstream ------------------------------------------
-
-    def get_session_id_by_ws(self, ws_id: str) -> str | None:
-        """Find the session_id associated with a workstream. Returns None if not found."""
-        ...
-
    # -- Channel routing -------------------------------------------------------

    def create_channel_route(
@@ -249,6 +230,89 @@ class StorageBackend(Protocol):
        """Remove a channel route. Returns True if existed."""
        ...

+    # -- Scheduled tasks -------------------------------------------------------
+
+    def create_scheduled_task(
+        self,
+        task_id: str,
+        name: str,
+        description: str,
+        schedule_type: str,
+        cron_expr: str,
+        at_time: str,
+        target_mode: str,
+        model: str,
+        initial_message: str,
+        auto_approve: bool,
+        auto_approve_tools: list[str],
+        created_by: str,
+        next_run: str,
+    ) -> None:
+        """Create a scheduled task. No-op if task_id already exists."""
+        ...
+
+    def get_scheduled_task(self, task_id: str) -> dict[str, Any] | None:
+        """Return scheduled task dict or None."""
+        ...
+
+    def list_scheduled_tasks(self) -> list[dict[str, Any]]:
+        """Return all scheduled tasks ordered by created DESC."""
+        ...
+
+    def update_scheduled_task(self, task_id: str, **fields: Any) -> bool:
+        """Update specified fields on a scheduled task. Returns True if found."""
+        ...
+
+    def delete_scheduled_task(self, task_id: str) -> bool:
+        """Delete a scheduled task and its run history. Returns True if found."""
+        ...
+
+    def list_due_tasks(self, now: str) -> list[dict[str, Any]]:
+        """Return enabled tasks whose next_run <= now, ordered by next_run."""
+        ...
+
+    def record_task_run(
+        self,
+        run_id: str,
+        task_id: str,
+        node_id: str,
+        ws_id: str,
+        correlation_id: str,
+        started: str,
+        status: str,
+        error: str,
+    ) -> None:
+        """Record a scheduled task execution."""
+        ...
+
+    def list_task_runs(self, task_id: str, limit: int = 50) -> list[dict[str, Any]]:
+        """List run history for a task, ordered by started DESC."""
+        ...
+
+    def prune_task_runs(self, retention_days: int = 90) -> int:
+        """Delete task runs older than retention_days. Returns count deleted."""
+        ...
+
+    # -- Service registry ------------------------------------------------------
+
+    def register_service(
+        self, service_type: str, service_id: str, url: str, metadata: str = "{}"
+    ) -> None:
+        """Register or update a service instance. Upserts by (service_type, service_id)."""
+        ...
+
+    def heartbeat_service(self, service_type: str, service_id: str) -> bool:
+        """Update last_heartbeat for a registered service. Returns False if not found."""
+        ...
+
+    def list_services(self, service_type: str, max_age_seconds: int = 120) -> list[dict[str, str]]:
+        """Return healthy services of a given type (heartbeat within max_age_seconds)."""
+        ...
+
+    def deregister_service(self, service_type: str, service_id: str) -> bool:
+        """Remove a service registration. Returns True if existed."""
+        ...
+
    # -- Lifecycle -------------------------------------------------------------

    def close(self) -> None:
@@ -22,7 +22,7 @@ conversations = sa.Table(
    "conversations",
    metadata,
    sa.Column("id", sa.Integer, primary_key=True, autoincrement=True),
-    sa.Column("session_id", sa.Text, nullable=False, index=True),
+    sa.Column("ws_id", sa.Text, nullable=False, index=True),
    sa.Column("timestamp", sa.Text, nullable=False),
    sa.Column("role", sa.Text, nullable=False),
    sa.Column("content", sa.Text),
@@ -32,32 +32,14 @@ conversations = sa.Table(
    sa.Column("provider_data", sa.Text),
 )

-sessions = sa.Table(
-    "sessions",
-    metadata,
-    sa.Column("session_id", sa.Text, primary_key=True),
-    sa.Column("alias", sa.Text, unique=True),
-    sa.Column("title", sa.Text),
-    sa.Column("node_id", sa.Text),
-    sa.Column("ws_id", sa.Text),
-    sa.Column("user_id", sa.Text),
-    sa.Column("created", sa.Text, nullable=False),
-    sa.Column("updated", sa.Text, nullable=False),
-)
-
-# Additional indexes on sessions (name-based to avoid duplication with SA's auto-index)
-sa.Index("idx_sessions_alias", sessions.c.alias)
-sa.Index("idx_sessions_updated", sessions.c.updated)
-sa.Index("idx_sessions_node_id", sessions.c.node_id)
-sa.Index("idx_sessions_ws_id", sessions.c.ws_id)
-sa.Index("idx_sessions_user_id", sessions.c.user_id)
-
 workstreams = sa.Table(
    "workstreams",
    metadata,
    sa.Column("ws_id", sa.Text, primary_key=True),
    sa.Column("node_id", sa.Text),
    sa.Column("user_id", sa.Text),
+    sa.Column("alias", sa.Text, unique=True),
+    sa.Column("title", sa.Text),
    sa.Column("name", sa.Text, nullable=False, server_default=""),
    sa.Column("state", sa.Text, nullable=False, server_default="idle"),
    sa.Column("created", sa.Text, nullable=False),
@@ -67,14 +49,15 @@ workstreams = sa.Table(
 sa.Index("idx_workstreams_node_id", workstreams.c.node_id)
 sa.Index("idx_workstreams_state", workstreams.c.state)
 sa.Index("idx_workstreams_user_id", workstreams.c.user_id)
+sa.Index("idx_workstreams_alias", workstreams.c.alias)

-session_config = sa.Table(
-    "session_config",
+workstream_config = sa.Table(
+    "workstream_config",
    metadata,
-    sa.Column("session_id", sa.Text, nullable=False),
+    sa.Column("ws_id", sa.Text, nullable=False),
    sa.Column("key", sa.Text, nullable=False),
    sa.Column("value", sa.Text),
-    sa.PrimaryKeyConstraint("session_id", "key"),
+    sa.PrimaryKeyConstraint("ws_id", "key"),
 )

 # ---------------------------------------------------------------------------
@@ -137,3 +120,66 @@ channel_routes = sa.Table(
 )

 sa.Index("idx_channel_routes_ws", channel_routes.c.ws_id)
+
+# ---------------------------------------------------------------------------
+# Scheduled task tables
+# ---------------------------------------------------------------------------
+
+scheduled_tasks = sa.Table(
+    "scheduled_tasks",
+    metadata,
+    sa.Column("task_id", sa.Text, primary_key=True),
+    sa.Column("name", sa.Text, nullable=False),
+    sa.Column("description", sa.Text, nullable=False, server_default=""),
+    sa.Column("schedule_type", sa.Text, nullable=False),  # "cron" or "at"
+    sa.Column("cron_expr", sa.Text, nullable=False, server_default=""),
+    sa.Column("at_time", sa.Text, nullable=False, server_default=""),  # ISO8601
+    sa.Column("target_mode", sa.Text, nullable=False, server_default="auto"),
+    sa.Column("model", sa.Text, nullable=False, server_default=""),
+    sa.Column("initial_message", sa.Text, nullable=False),
+    sa.Column("auto_approve", sa.Integer, nullable=False, server_default="0"),
+    sa.Column("auto_approve_tools", sa.Text, nullable=False, server_default=""),
+    sa.Column("enabled", sa.Integer, nullable=False, server_default="1"),
+    sa.Column("created_by", sa.Text, nullable=False, server_default=""),
+    sa.Column("last_run", sa.Text),
+    sa.Column("next_run", sa.Text),
+    sa.Column("created", sa.Text, nullable=False),
+    sa.Column("updated", sa.Text, nullable=False),
+)
+
+sa.Index("idx_scheduled_tasks_enabled", scheduled_tasks.c.enabled)
+sa.Index("idx_scheduled_tasks_next_run", scheduled_tasks.c.next_run)
+
+scheduled_task_runs = sa.Table(
+    "scheduled_task_runs",
+    metadata,
+    sa.Column("run_id", sa.Text, primary_key=True),
+    sa.Column("task_id", sa.Text, nullable=False),
+    sa.Column("node_id", sa.Text, nullable=False, server_default=""),
+    sa.Column("ws_id", sa.Text, nullable=False, server_default=""),
+    sa.Column("correlation_id", sa.Text, nullable=False, server_default=""),
+    sa.Column("started", sa.Text, nullable=False),
+    sa.Column("status", sa.Text, nullable=False, server_default="dispatched"),
+    sa.Column("error", sa.Text, nullable=False, server_default=""),
+)
+
+sa.Index("idx_scheduled_task_runs_task_id", scheduled_task_runs.c.task_id)
+sa.Index("idx_scheduled_task_runs_started", scheduled_task_runs.c.started)
+
+# ---------------------------------------------------------------------------
+# Service registry
+# ---------------------------------------------------------------------------
+
+services = sa.Table(
+    "services",
+    metadata,
+    sa.Column("service_type", sa.Text, nullable=False),
+    sa.Column("service_id", sa.Text, nullable=False),
+    sa.Column("url", sa.Text, nullable=False),
+    sa.Column("metadata", sa.Text, nullable=False, server_default="{}"),
+    sa.Column("last_heartbeat", sa.Text, nullable=False),
+    sa.Column("created", sa.Text, nullable=False),
+    sa.PrimaryKeyConstraint("service_type", "service_id"),
+)
+
+sa.Index("idx_services_type_heartbeat", services.c.service_type, services.c.last_heartbeat)
@@ -15,9 +15,8 @@ from turnstone.core.storage._schema import (
    conversations,
    memories,
    metadata,
-    session_config,
-    sessions,
    users,
+    workstream_config,
    workstreams,
 )

@@ -82,35 +81,11 @@ class SQLiteBackend:
            except Exception:
                self._fts5_available = False

-    # -- Core session operations -----------------------------------------------
-
-    def register_session(
-        self,
-        session_id: str,
-        title: str | None = None,
-        node_id: str | None = None,
-        ws_id: str | None = None,
-        user_id: str | None = None,
-    ) -> None:
-        now = datetime.now(UTC).strftime("%Y-%m-%dT%H:%M:%S")
-        with self._engine.connect() as conn:
-            conn.execute(
-                sa.insert(sessions).prefix_with("OR IGNORE"),
-                {
-                    "session_id": session_id,
-                    "title": title,
-                    "node_id": node_id,
-                    "ws_id": ws_id,
-                    "user_id": user_id,
-                    "created": now,
-                    "updated": now,
-                },
-            )
-            conn.commit()
+    # -- Core conversation operations ------------------------------------------

    def save_message(
        self,
-        session_id: str,
+        ws_id: str,
        role: str,
        content: str | None,
        tool_name: str | None = None,
@@ -123,7 +98,7 @@ class SQLiteBackend:
            result = conn.execute(
                sa.insert(conversations),
                {
-                    "session_id": session_id,
+                    "ws_id": ws_id,
                    "timestamp": now,
                    "role": role,
                    "content": content,
@@ -145,13 +120,13 @@ class SQLiteBackend:
                    )
                except Exception:
                    self._fts5_available = False
-            # Bump session updated timestamp
+            # Bump workstream updated timestamp
            conn.execute(
-                sa.update(sessions).where(sessions.c.session_id == session_id).values(updated=now)
+                sa.update(workstreams).where(workstreams.c.ws_id == ws_id).values(updated=now)
            )
            conn.commit()

-    def load_session_messages(self, session_id: str) -> list[dict[str, Any]]:
+    def load_messages(self, ws_id: str) -> list[dict[str, Any]]:
        with self._engine.connect() as conn:
            rows = conn.execute(
                sa.select(
@@ -162,52 +137,44 @@ class SQLiteBackend:
                    conversations.c.tool_call_id,
                    conversations.c.provider_data,
                )
-                .where(conversations.c.session_id == session_id)
+                .where(conversations.c.ws_id == ws_id)
                .order_by(conversations.c.id)
            ).fetchall()

-        return _reconstruct_messages(list(rows), session_id)
+        return _reconstruct_messages(list(rows), ws_id)

-    # -- Session management ----------------------------------------------------
+    # -- Workstream management -------------------------------------------------

-    def list_sessions(self, limit: int = 20) -> list[Any]:
+    def list_workstreams_with_history(self, limit: int = 20) -> list[Any]:
        with self._engine.connect() as conn:
            return list(
                conn.execute(
                    sa.text(
-                        "SELECT s.session_id, s.alias, s.title, s.created, s.updated, "
+                        "SELECT w.ws_id, w.alias, w.title, w.created, w.updated, "
                        "(SELECT COUNT(*) FROM conversations c "
-                        " WHERE c.session_id = s.session_id), "
-                        "s.node_id, s.ws_id "
-                        "FROM sessions s "
+                        " WHERE c.ws_id = w.ws_id), "
+                        "w.node_id "
+                        "FROM workstreams w "
                        "WHERE EXISTS "
-                        "  (SELECT 1 FROM conversations c WHERE c.session_id = s.session_id) "
-                        "ORDER BY s.updated DESC LIMIT :limit"
+                        "  (SELECT 1 FROM conversations c WHERE c.ws_id = w.ws_id) "
+                        "ORDER BY w.updated DESC LIMIT :limit"
                    ),
                    {"limit": limit},
                ).fetchall()
            )

-    def delete_session(self, session_id: str) -> bool:
-        with self._engine.connect() as conn:
-            conn.execute(sa.delete(conversations).where(conversations.c.session_id == session_id))
-            conn.execute(sa.delete(session_config).where(session_config.c.session_id == session_id))
-            conn.execute(sa.delete(sessions).where(sessions.c.session_id == session_id))
-            conn.commit()
-            return True
-
-    def prune_sessions(self, retention_days: int = 90) -> tuple[int, int]:
+    def prune_workstreams(self, retention_days: int = 90) -> tuple[int, int]:
        orphans = stale = 0
        with self._engine.connect() as conn:
-            # 1. Remove sessions with no messages
+            # 1. Remove workstreams with no messages
            orphan_ids = [
                row[0]
                for row in conn.execute(
                    sa.text(
-                        "SELECT session_id FROM sessions "
+                        "SELECT ws_id FROM workstreams "
                        "WHERE NOT EXISTS "
                        "  (SELECT 1 FROM conversations c "
-                        "   WHERE c.session_id = sessions.session_id)"
+                        "   WHERE c.ws_id = workstreams.ws_id)"
                    )
                ).fetchall()
            ]
@@ -215,16 +182,16 @@ class SQLiteBackend:
                placeholders = ",".join([":p" + str(i) for i in range(len(orphan_ids))])
                params = {f"p{i}": oid for i, oid in enumerate(orphan_ids)}
                conn.execute(
-                    sa.text(f"DELETE FROM session_config WHERE session_id IN ({placeholders})"),
+                    sa.text(f"DELETE FROM workstream_config WHERE ws_id IN ({placeholders})"),
                    params,
                )
                result = conn.execute(
-                    sa.text(f"DELETE FROM sessions WHERE session_id IN ({placeholders})"),
+                    sa.text(f"DELETE FROM workstreams WHERE ws_id IN ({placeholders})"),
                    params,
                )
                orphans = result.rowcount

-            # 2. Remove old unnamed sessions
+            # 2. Remove old unnamed workstreams
            if retention_days > 0:
                cutoff = (datetime.now(UTC) - timedelta(days=retention_days)).strftime(
                    "%Y-%m-%dT%H:%M:%S"
@@ -233,7 +200,7 @@ class SQLiteBackend:
                    row[0]
                    for row in conn.execute(
                        sa.text(
-                            "SELECT session_id FROM sessions "
+                            "SELECT ws_id FROM workstreams "
                            "WHERE alias IS NULL AND updated < :cutoff"
                        ),
                        {"cutoff": cutoff},
@@ -243,11 +210,15 @@ class SQLiteBackend:
                    placeholders = ",".join([":p" + str(i) for i in range(len(stale_ids))])
                    params = {f"p{i}": sid for i, sid in enumerate(stale_ids)}
                    conn.execute(
-                        sa.text(f"DELETE FROM session_config WHERE session_id IN ({placeholders})"),
+                        sa.text(f"DELETE FROM workstream_config WHERE ws_id IN ({placeholders})"),
+                        params,
+                    )
+                    conn.execute(
+                        sa.text(f"DELETE FROM conversations WHERE ws_id IN ({placeholders})"),
                        params,
                    )
                    result = conn.execute(
-                        sa.text(f"DELETE FROM sessions WHERE session_id IN ({placeholders})"),
+                        sa.text(f"DELETE FROM workstreams WHERE ws_id IN ({placeholders})"),
                        params,
                    )
                    stale = result.rowcount
@@ -255,94 +226,71 @@ class SQLiteBackend:
            conn.commit()
        return (orphans, stale)

-    def resolve_session(self, alias_or_id: str) -> str | None:
+    def resolve_workstream(self, alias_or_id: str) -> str | None:
        with self._engine.connect() as conn:
            # 1. Exact alias match
            row = conn.execute(
-                sa.select(sessions.c.session_id).where(sessions.c.alias == alias_or_id)
+                sa.select(workstreams.c.ws_id).where(workstreams.c.alias == alias_or_id)
            ).fetchone()
            if row:
                return str(row[0])
-            # 2. Exact session_id match
+            # 2. Exact ws_id match
            row = conn.execute(
-                sa.select(sessions.c.session_id).where(sessions.c.session_id == alias_or_id)
+                sa.select(workstreams.c.ws_id).where(workstreams.c.ws_id == alias_or_id)
            ).fetchone()
            if row:
                return str(row[0])
-            # 3. Session_id prefix match
+            # 3. ws_id prefix match
            rows = conn.execute(
-                sa.select(sessions.c.session_id).where(
-                    sessions.c.session_id.like(alias_or_id + "%")
-                )
+                sa.select(workstreams.c.ws_id).where(workstreams.c.ws_id.like(alias_or_id + "%"))
            ).fetchall()
            if len(rows) == 1:
                return str(rows[0][0])
-            # 4. Legacy: check conversations table
-            row = conn.execute(
-                sa.text(
-                    "SELECT DISTINCT session_id FROM conversations WHERE session_id = :sid LIMIT 1"
-                ),
-                {"sid": alias_or_id},
-            ).fetchone()
-            if row:
-                # Auto-register legacy session
-                conn.execute(
-                    sa.text(
-                        "INSERT OR IGNORE INTO sessions "
-                        "(session_id, created, updated) VALUES ("
-                        ":sid, "
-                        "(SELECT MIN(timestamp) FROM conversations WHERE session_id = :sid), "
-                        "(SELECT MAX(timestamp) FROM conversations WHERE session_id = :sid))"
-                    ),
-                    {"sid": row[0]},
-                )
-                conn.commit()
-                return str(row[0])
            return None

-    # -- Session config --------------------------------------------------------
+    # -- Workstream config -----------------------------------------------------

-    def save_session_config(self, session_id: str, config: dict[str, str]) -> None:
+    def save_workstream_config(self, ws_id: str, config: dict[str, str]) -> None:
        with self._engine.connect() as conn:
            for key, value in config.items():
                conn.execute(
                    sa.text(
-                        "INSERT OR REPLACE INTO session_config "
-                        "(session_id, key, value) VALUES (:sid, :key, :value)"
+                        "INSERT OR REPLACE INTO workstream_config "
+                        "(ws_id, key, value) VALUES (:wid, :key, :value)"
                    ),
-                    {"sid": session_id, "key": key, "value": value},
+                    {"wid": ws_id, "key": key, "value": value},
                )
            conn.commit()

-    def load_session_config(self, session_id: str) -> dict[str, str]:
+    def load_workstream_config(self, ws_id: str) -> dict[str, str]:
        with self._engine.connect() as conn:
            rows = conn.execute(
-                sa.select(session_config.c.key, session_config.c.value).where(
-                    session_config.c.session_id == session_id
+                sa.select(workstream_config.c.key, workstream_config.c.value).where(
+                    workstream_config.c.ws_id == ws_id
                )
            ).fetchall()
            return {row[0]: row[1] for row in rows}

-    # -- Session metadata ------------------------------------------------------
+    # -- Workstream metadata ---------------------------------------------------

-    def set_session_alias(self, session_id: str, alias: str) -> bool:
+    def set_workstream_alias(self, ws_id: str, alias: str) -> bool:
        with self._engine.connect() as conn:
            existing = conn.execute(
-                sa.select(sessions.c.session_id).where(sessions.c.alias == alias)
+                sa.select(workstreams.c.ws_id).where(workstreams.c.alias == alias)
            ).fetchone()
-            if existing and existing[0] != session_id:
+            if existing and existing[0] != ws_id:
                return False
            conn.execute(
-                sa.update(sessions).where(sessions.c.session_id == session_id).values(alias=alias)
+                sa.update(workstreams).where(workstreams.c.ws_id == ws_id).values(alias=alias)
            )
            conn.commit()
            return True

-    def get_session_name(self, session_id: str) -> str | None:
+    def get_workstream_display_name(self, ws_id: str) -> str | None:
        with self._engine.connect() as conn:
            row = conn.execute(
-                sa.select(sessions.c.alias, sessions.c.title).where(
-                    sessions.c.session_id == session_id
+                sa.select(workstreams.c.alias, workstreams.c.title).where(
+                    workstreams.c.ws_id == ws_id
                )
            ).fetchone()
            if row:
@@ -350,10 +298,10 @@ class SQLiteBackend:
                return str(value) if value is not None else None
        return None

-    def update_session_title(self, session_id: str, title: str) -> None:
+    def update_workstream_title(self, ws_id: str, title: str) -> None:
        with self._engine.connect() as conn:
            conn.execute(
-                sa.update(sessions).where(sessions.c.session_id == session_id).values(title=title)
+                sa.update(workstreams).where(workstreams.c.ws_id == ws_id).values(title=title)
            )
            conn.commit()

@@ -428,6 +376,8 @@ class SQLiteBackend:
        name: str = "",
        state: str = "idle",
        user_id: str | None = None,
+        alias: str | None = None,
+        title: str | None = None,
    ) -> None:
        now = datetime.now(UTC).strftime("%Y-%m-%dT%H:%M:%S")
        with self._engine.connect() as conn:
@@ -437,6 +387,8 @@ class SQLiteBackend:
                    "ws_id": ws_id,
                    "node_id": node_id,
                    "user_id": user_id,
+                    "alias": alias,
+                    "title": title,
                    "name": name,
                    "state": state,
                    "created": now,
@@ -467,6 +419,8 @@ class SQLiteBackend:

    def delete_workstream(self, ws_id: str) -> bool:
        with self._engine.connect() as conn:
+            conn.execute(sa.delete(conversations).where(conversations.c.ws_id == ws_id))
+            conn.execute(sa.delete(workstream_config).where(workstream_config.c.ws_id == ws_id))
            result = conn.execute(sa.delete(workstreams).where(workstreams.c.ws_id == ws_id))
            conn.commit()
            return result.rowcount > 0
@@ -500,7 +454,7 @@ class SQLiteBackend:
                return list(
                    conn.execute(
                        sa.text(
-                            "SELECT c.timestamp, c.session_id, c.role, c.content, c.tool_name "
+                            "SELECT c.timestamp, c.ws_id, c.role, c.content, c.tool_name "
                            "FROM conversations_fts f "
                            "JOIN conversations c ON c.id = f.rowid "
                            "WHERE conversations_fts MATCH :query "
@@ -512,7 +466,7 @@ class SQLiteBackend:
            return list(
                conn.execute(
                    sa.text(
-                        "SELECT timestamp, session_id, role, content, tool_name "
+                        "SELECT timestamp, ws_id, role, content, tool_name "
                        "FROM conversations WHERE content LIKE :pattern ESCAPE '\\' "
                        "ORDER BY timestamp DESC LIMIT :limit"
                    ),
@@ -526,7 +480,7 @@ class SQLiteBackend:
            return list(
                conn.execute(
                    sa.text(
-                        "SELECT timestamp, session_id, role, content, tool_name "
+                        "SELECT timestamp, ws_id, role, content, tool_name "
                        "FROM conversations ORDER BY timestamp DESC LIMIT :limit"
                    ),
                    {"limit": capped},
@@ -732,15 +686,6 @@ class SQLiteBackend:
            conn.commit()
            return result.rowcount > 0

-    # -- Session lookup by workstream ------------------------------------------
-
-    def get_session_id_by_ws(self, ws_id: str) -> str | None:
-        with self._engine.connect() as conn:
-            row = conn.execute(
-                sa.select(sessions.c.session_id).where(sessions.c.ws_id == ws_id)
-            ).fetchone()
-            return str(row[0]) if row else None
-
    # -- Channel user mapping ---------------------------------------------------

    def create_channel_user(self, channel_type: str, channel_user_id: str, user_id: str) -> None:
@@ -929,13 +874,273 @@ class SQLiteBackend:
            conn.commit()
            return result.rowcount > 0

+    # -- Scheduled tasks -------------------------------------------------------
+
+    def create_scheduled_task(
+        self,
+        task_id: str,
+        name: str,
+        description: str,
+        schedule_type: str,
+        cron_expr: str,
+        at_time: str,
+        target_mode: str,
+        model: str,
+        initial_message: str,
+        auto_approve: bool,
+        auto_approve_tools: list[str],
+        created_by: str,
+        next_run: str,
+    ) -> None:
+        from turnstone.core.storage._schema import scheduled_tasks
+
+        now = datetime.now(UTC).strftime("%Y-%m-%dT%H:%M:%S")
+        with self._engine.connect() as conn:
+            conn.execute(
+                sa.insert(scheduled_tasks).prefix_with("OR IGNORE"),
+                {
+                    "task_id": task_id,
+                    "name": name,
+                    "description": description,
+                    "schedule_type": schedule_type,
+                    "cron_expr": cron_expr,
+                    "at_time": at_time,
+                    "target_mode": target_mode,
+                    "model": model,
+                    "initial_message": initial_message,
+                    "auto_approve": 1 if auto_approve else 0,
+                    "auto_approve_tools": ",".join(auto_approve_tools),
+                    "enabled": 1,
+                    "created_by": created_by,
+                    "next_run": next_run,
+                    "created": now,
+                    "updated": now,
+                },
+            )
+            conn.commit()
+
+    def get_scheduled_task(self, task_id: str) -> dict[str, Any] | None:
+        from turnstone.core.storage._schema import scheduled_tasks
+
+        with self._engine.connect() as conn:
+            row = conn.execute(
+                sa.select(scheduled_tasks).where(scheduled_tasks.c.task_id == task_id)
+            ).fetchone()
+            if row is None:
+                return None
+            return dict(row._mapping)
+
+    def list_scheduled_tasks(self) -> list[dict[str, Any]]:
+        from turnstone.core.storage._schema import scheduled_tasks
+
+        with self._engine.connect() as conn:
+            rows = conn.execute(
+                sa.select(scheduled_tasks).order_by(scheduled_tasks.c.created.desc())
+            ).fetchall()
+            return [dict(r._mapping) for r in rows]
+
+    _UPDATABLE_TASK_FIELDS = frozenset(
+        {
+            "name",
+            "description",
+            "schedule_type",
+            "cron_expr",
+            "at_time",
+            "target_mode",
+            "model",
+            "initial_message",
+            "auto_approve",
+            "auto_approve_tools",
+            "enabled",
+            "last_run",
+            "next_run",
+            "updated",
+        }
+    )
+
+    def update_scheduled_task(self, task_id: str, **fields: Any) -> bool:
+        from turnstone.core.storage._schema import scheduled_tasks
+
+        fields = {k: v for k, v in fields.items() if k in self._UPDATABLE_TASK_FIELDS}
+        fields["updated"] = datetime.now(UTC).strftime("%Y-%m-%dT%H:%M:%S")
+        # Normalize boolean → int for auto_approve
+        if "auto_approve" in fields:
+            fields["auto_approve"] = 1 if fields["auto_approve"] else 0
+        if "auto_approve_tools" in fields and isinstance(fields["auto_approve_tools"], list):
+            fields["auto_approve_tools"] = ",".join(fields["auto_approve_tools"])
+        if "enabled" in fields:
+            fields["enabled"] = 1 if fields["enabled"] else 0
+        with self._engine.connect() as conn:
+            result = conn.execute(
+                sa.update(scheduled_tasks)
+                .where(scheduled_tasks.c.task_id == task_id)
+                .values(**fields)
+            )
+            conn.commit()
+            return result.rowcount > 0
+
+    def delete_scheduled_task(self, task_id: str) -> bool:
+        from turnstone.core.storage._schema import scheduled_task_runs, scheduled_tasks
+
+        with self._engine.connect() as conn:
+            conn.execute(
+                sa.delete(scheduled_task_runs).where(scheduled_task_runs.c.task_id == task_id)
+            )
+            result = conn.execute(
+                sa.delete(scheduled_tasks).where(scheduled_tasks.c.task_id == task_id)
+            )
+            conn.commit()
+            return result.rowcount > 0
+
+    def list_due_tasks(self, now: str) -> list[dict[str, Any]]:
+        from turnstone.core.storage._schema import scheduled_tasks
+
+        with self._engine.connect() as conn:
+            rows = conn.execute(
+                sa.select(scheduled_tasks)
+                .where(
+                    (scheduled_tasks.c.enabled == 1)
+                    & (scheduled_tasks.c.next_run <= now)
+                    & (scheduled_tasks.c.next_run != "")
+                )
+                .order_by(scheduled_tasks.c.next_run)
+                .limit(100)
+            ).fetchall()
+            return [dict(r._mapping) for r in rows]
+
+    def record_task_run(
+        self,
+        run_id: str,
+        task_id: str,
+        node_id: str,
+        ws_id: str,
+        correlation_id: str,
+        started: str,
+        status: str,
+        error: str,
+    ) -> None:
+        from turnstone.core.storage._schema import scheduled_task_runs
+
+        with self._engine.connect() as conn:
+            conn.execute(
+                sa.insert(scheduled_task_runs),
+                {
+                    "run_id": run_id,
+                    "task_id": task_id,
+                    "node_id": node_id,
+                    "ws_id": ws_id,
+                    "correlation_id": correlation_id,
+                    "started": started,
+                    "status": status,
+                    "error": error,
+                },
+            )
+            conn.commit()
+
+    def list_task_runs(self, task_id: str, limit: int = 50) -> list[dict[str, Any]]:
+        from turnstone.core.storage._schema import scheduled_task_runs
+
+        with self._engine.connect() as conn:
+            rows = conn.execute(
+                sa.select(scheduled_task_runs)
+                .where(scheduled_task_runs.c.task_id == task_id)
+                .order_by(scheduled_task_runs.c.started.desc())
+                .limit(limit)
+            ).fetchall()
+            return [dict(r._mapping) for r in rows]
+
+    def prune_task_runs(self, retention_days: int = 90) -> int:
+        from datetime import timedelta
+
+        from turnstone.core.storage._schema import scheduled_task_runs
+
+        cutoff = (datetime.now(UTC) - timedelta(days=retention_days)).strftime("%Y-%m-%dT%H:%M:%S")
+        with self._engine.connect() as conn:
+            result = conn.execute(
+                sa.delete(scheduled_task_runs).where(scheduled_task_runs.c.started < cutoff)
+            )
+            conn.commit()
+            return result.rowcount
+
+    # -- Service registry ------------------------------------------------------
+
+    def register_service(
+        self, service_type: str, service_id: str, url: str, metadata: str = "{}"
+    ) -> None:
+        from sqlalchemy.dialects.sqlite import insert as sqlite_insert
+
+        from turnstone.core.storage._schema import services
+
+        now = datetime.now(UTC).strftime("%Y-%m-%dT%H:%M:%S")
+        stmt = sqlite_insert(services).values(
+            service_type=service_type,
+            service_id=service_id,
+            url=url,
+            metadata=metadata,
+            last_heartbeat=now,
+            created=now,
+        )
+        stmt = stmt.on_conflict_do_update(
+            index_elements=["service_type", "service_id"],
+            set_={"url": url, "metadata": metadata, "last_heartbeat": now},
+        )
+        with self._engine.connect() as conn:
+            conn.execute(stmt)
+            conn.commit()
+
+    def heartbeat_service(self, service_type: str, service_id: str) -> bool:
+        from turnstone.core.storage._schema import services
+
+        now = datetime.now(UTC).strftime("%Y-%m-%dT%H:%M:%S")
+        with self._engine.connect() as conn:
+            result = conn.execute(
+                sa.update(services)
+                .where(
+                    (services.c.service_type == service_type)
+                    & (services.c.service_id == service_id)
+                )
+                .values(last_heartbeat=now)
+            )
+            conn.commit()
+            return result.rowcount > 0
+
+    def list_services(self, service_type: str, max_age_seconds: int = 120) -> list[dict[str, str]]:
+        from turnstone.core.storage._schema import services
+
+        cutoff = (datetime.now(UTC) - timedelta(seconds=max_age_seconds)).strftime(
+            "%Y-%m-%dT%H:%M:%S"
+        )
+        with self._engine.connect() as conn:
+            rows = conn.execute(
+                sa.select(services)
+                .where(
+                    (services.c.service_type == service_type)
+                    & (services.c.last_heartbeat >= cutoff)
+                )
+                .order_by(services.c.last_heartbeat.desc())
+            ).fetchall()
+            return [dict(r._mapping) for r in rows]
+
+    def deregister_service(self, service_type: str, service_id: str) -> bool:
+        from turnstone.core.storage._schema import services
+
+        with self._engine.connect() as conn:
+            result = conn.execute(
+                sa.delete(services).where(
+                    (services.c.service_type == service_type)
+                    & (services.c.service_id == service_id)
+                )
+            )
+            conn.commit()
+            return result.rowcount > 0
+
    # -- Lifecycle -------------------------------------------------------------

    def close(self) -> None:
        self._engine.dispose()


-def _reconstruct_messages(rows: list[Any], session_id: str) -> list[dict[str, Any]]:
+def _reconstruct_messages(rows: list[Any], ws_id: str) -> list[dict[str, Any]]:
    """Reconstruct OpenAI message format from stored conversation rows.

    Handles tool_call / tool_result grouping and incomplete turn repair.
@@ -973,7 +1178,7 @@ def _reconstruct_messages(rows: list[Any], session_id: str) -> list[dict[str, An

            while i < len(rows) and rows[i][0] == "tool_call":
                _, _, tn, ta, stored_tc_id, _ = rows[i]
-                call_id = stored_tc_id or f"call_{session_id}_{i}"
+                call_id = stored_tc_id or f"call_{ws_id}_{i}"
                assistant_msg["tool_calls"].append(
                    {
                        "id": call_id,
@@ -0,0 +1,62 @@
+"""Scheduled tasks and run history tables.
+
+Revision ID: 004
+Revises: 003
+Create Date: 2026-03-05
+"""
+
+import sqlalchemy as sa
+from alembic import op
+
+revision = "004"
+down_revision = "003"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    op.create_table(
+        "scheduled_tasks",
+        sa.Column("task_id", sa.Text, primary_key=True),
+        sa.Column("name", sa.Text, nullable=False),
+        sa.Column("description", sa.Text, nullable=False, server_default=""),
+        sa.Column("schedule_type", sa.Text, nullable=False),
+        sa.Column("cron_expr", sa.Text, nullable=False, server_default=""),
+        sa.Column("at_time", sa.Text, nullable=False, server_default=""),
+        sa.Column("target_mode", sa.Text, nullable=False, server_default="auto"),
+        sa.Column("model", sa.Text, nullable=False, server_default=""),
+        sa.Column("initial_message", sa.Text, nullable=False),
+        sa.Column("auto_approve", sa.Integer, nullable=False, server_default="0"),
+        sa.Column("auto_approve_tools", sa.Text, nullable=False, server_default=""),
+        sa.Column("enabled", sa.Integer, nullable=False, server_default="1"),
+        sa.Column("created_by", sa.Text, nullable=False, server_default=""),
+        sa.Column("last_run", sa.Text),
+        sa.Column("next_run", sa.Text),
+        sa.Column("created", sa.Text, nullable=False),
+        sa.Column("updated", sa.Text, nullable=False),
+    )
+    op.create_index("idx_scheduled_tasks_enabled", "scheduled_tasks", ["enabled"])
+    op.create_index("idx_scheduled_tasks_next_run", "scheduled_tasks", ["next_run"])
+
+    op.create_table(
+        "scheduled_task_runs",
+        sa.Column("run_id", sa.Text, primary_key=True),
+        sa.Column("task_id", sa.Text, nullable=False),
+        sa.Column("node_id", sa.Text, nullable=False, server_default=""),
+        sa.Column("ws_id", sa.Text, nullable=False, server_default=""),
+        sa.Column("correlation_id", sa.Text, nullable=False, server_default=""),
+        sa.Column("started", sa.Text, nullable=False),
+        sa.Column("status", sa.Text, nullable=False, server_default="dispatched"),
+        sa.Column("error", sa.Text, nullable=False, server_default=""),
+    )
+    op.create_index("idx_scheduled_task_runs_task_id", "scheduled_task_runs", ["task_id"])
+    op.create_index("idx_scheduled_task_runs_started", "scheduled_task_runs", ["started"])
+
+
+def downgrade() -> None:
+    op.drop_index("idx_scheduled_task_runs_started", "scheduled_task_runs")
+    op.drop_index("idx_scheduled_task_runs_task_id", "scheduled_task_runs")
+    op.drop_table("scheduled_task_runs")
+    op.drop_index("idx_scheduled_tasks_next_run", "scheduled_tasks")
+    op.drop_index("idx_scheduled_tasks_enabled", "scheduled_tasks")
+    op.drop_table("scheduled_tasks")
@@ -0,0 +1,33 @@
+"""Service registry table.
+
+Revision ID: 005
+Revises: 004
+Create Date: 2026-03-05
+"""
+
+import sqlalchemy as sa
+from alembic import op
+
+revision = "005"
+down_revision = "004"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    op.create_table(
+        "services",
+        sa.Column("service_type", sa.Text, nullable=False),
+        sa.Column("service_id", sa.Text, nullable=False),
+        sa.Column("url", sa.Text, nullable=False),
+        sa.Column("metadata", sa.Text, nullable=False, server_default="{}"),
+        sa.Column("last_heartbeat", sa.Text, nullable=False),
+        sa.Column("created", sa.Text, nullable=False),
+        sa.PrimaryKeyConstraint("service_type", "service_id"),
+    )
+    op.create_index("idx_services_type_heartbeat", "services", ["service_type", "last_heartbeat"])
+
+
+def downgrade() -> None:
+    op.drop_index("idx_services_type_heartbeat", "services")
+    op.drop_table("services")
@@ -0,0 +1,131 @@
+"""Normalize session_id into ws_id — merge sessions table into workstreams.
+
+Conversations and config are now keyed by ws_id (workstream identity) instead
+of a separate session_id.  The sessions table is dropped; its alias/title
+columns move to workstreams.
+
+Revision ID: 006
+Revises: 005
+Create Date: 2026-03-07
+"""
+
+import sqlalchemy as sa
+from alembic import op
+
+revision = "006"
+down_revision = "005"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    # 1. Add alias and title columns to workstreams.
+    op.add_column("workstreams", sa.Column("alias", sa.Text))
+    op.add_column("workstreams", sa.Column("title", sa.Text))
+    op.create_index("idx_workstreams_alias", "workstreams", ["alias"], unique=True)
+
+    conn = op.get_bind()
+
+    # 2. Copy alias/title from sessions → workstreams (for rows that have a ws_id).
+    conn.execute(
+        sa.text(
+            "UPDATE workstreams SET "
+            "  alias = (SELECT s.alias FROM sessions s WHERE s.ws_id = workstreams.ws_id), "
+            "  title = (SELECT s.title FROM sessions s WHERE s.ws_id = workstreams.ws_id) "
+            "WHERE EXISTS (SELECT 1 FROM sessions s WHERE s.ws_id = workstreams.ws_id)"
+        )
+    )
+
+    # 3. Create workstream rows for sessions that have a ws_id but no
+    #    corresponding workstream row yet.
+    #    The NOT EXISTS guard makes this safe on both SQLite and PostgreSQL.
+    conn.execute(
+        sa.text(
+            "INSERT INTO workstreams "
+            "(ws_id, node_id, alias, title, state, created, updated) "
+            "SELECT s.ws_id, s.node_id, s.alias, s.title, 'closed', s.created, s.updated "
+            "FROM sessions s "
+            "WHERE s.ws_id IS NOT NULL AND s.ws_id != '' "
+            "  AND NOT EXISTS (SELECT 1 FROM workstreams w WHERE w.ws_id = s.ws_id)"
+        )
+    )
+
+    # 4. Create workstream rows for sessions WITHOUT a ws_id
+    #    (use session_id as ws_id).
+    conn.execute(
+        sa.text(
+            "INSERT INTO workstreams "
+            "(ws_id, node_id, alias, title, state, created, updated) "
+            "SELECT s.session_id, s.node_id, s.alias, s.title, 'closed', s.created, s.updated "
+            "FROM sessions s "
+            "WHERE (s.ws_id IS NULL OR s.ws_id = '') "
+            "  AND NOT EXISTS (SELECT 1 FROM workstreams w WHERE w.ws_id = s.session_id)"
+        )
+    )
+
+    # 5. Rename conversations.session_id → conversations.ws_id and remap values.
+    #    For sessions with ws_id: map session_id → ws_id.
+    #    For sessions without ws_id: session_id stays (used as ws_id).
+    op.alter_column("conversations", "session_id", new_column_name="ws_id")
+
+    conn.execute(
+        sa.text(
+            "UPDATE conversations SET ws_id = ("
+            "  SELECT COALESCE(NULLIF(s.ws_id, ''), s.session_id) "
+            "  FROM sessions s WHERE s.session_id = conversations.ws_id"
+            ") "
+            "WHERE EXISTS ("
+            "  SELECT 1 FROM sessions s WHERE s.session_id = conversations.ws_id"
+            ")"
+        )
+    )
+
+    # 6. Rename session_config → workstream_config with ws_id column.
+    op.rename_table("session_config", "workstream_config")
+    op.alter_column("workstream_config", "session_id", new_column_name="ws_id")
+
+    conn.execute(
+        sa.text(
+            "UPDATE workstream_config SET ws_id = ("
+            "  SELECT COALESCE(NULLIF(s.ws_id, ''), s.session_id) "
+            "  FROM sessions s WHERE s.session_id = workstream_config.ws_id"
+            ") "
+            "WHERE EXISTS ("
+            "  SELECT 1 FROM sessions s WHERE s.session_id = workstream_config.ws_id"
+            ")"
+        )
+    )
+
+    # 7. Drop the sessions table.
+    op.drop_table("sessions")
+
+
+def downgrade() -> None:
+    # Recreate the sessions table.
+    op.create_table(
+        "sessions",
+        sa.Column("session_id", sa.Text, primary_key=True),
+        sa.Column("alias", sa.Text, unique=True),
+        sa.Column("title", sa.Text),
+        sa.Column("node_id", sa.Text),
+        sa.Column("ws_id", sa.Text),
+        sa.Column("user_id", sa.Text),
+        sa.Column("created", sa.Text, nullable=False),
+        sa.Column("updated", sa.Text, nullable=False),
+    )
+    op.create_index("idx_sessions_alias", "sessions", ["alias"])
+    op.create_index("idx_sessions_updated", "sessions", ["updated"])
+    op.create_index("idx_sessions_node_id", "sessions", ["node_id"])
+    op.create_index("idx_sessions_ws_id", "sessions", ["ws_id"])
+
+    # Reverse config table rename.
+    op.alter_column("workstream_config", "ws_id", new_column_name="session_id")
+    op.rename_table("workstream_config", "session_config")
+
+    # Reverse conversations column rename.
+    op.alter_column("conversations", "ws_id", new_column_name="session_id")
+
+    # Drop alias/title from workstreams.
+    op.drop_index("idx_workstreams_alias", "workstreams")
+    op.drop_column("workstreams", "title")
+    op.drop_column("workstreams", "alias")
@@ -1,4 +1,4 @@
-"""Workstream manager — concurrent independent chat sessions.
+"""Workstream manager — concurrent independent conversations.

 A workstream is an independent conversation with its own ChatSession and UI
 adapter.  The WorkstreamManager coordinates multiple workstreams, tracks their
@@ -74,9 +74,10 @@ class WorkstreamManager:
        Args:
            session_factory: callable(ui, model_alias, ws_id) -> ChatSession.
                Captures shared config (registry, temperature, …) so the
-                manager can create sessions without knowing those details.
-                *model_alias* selects a model from the registry (None = default).
-                *ws_id* links the session to its workstream in storage.
+                manager can create ChatSession instances without knowing
+                those details.  *model_alias* selects a model from the
+                registry (None = default).  *ws_id* is the persistent
+                identity used for all storage operations.
            max_workstreams: Maximum number of concurrent workstreams.  When at
                capacity, ``create()`` will auto-evict the oldest IDLE
                workstream before raising.
@@ -125,7 +126,7 @@ class WorkstreamManager:
            model: Optional model alias from the registry.  ``None`` uses the
                default model.
        """
-        # Fast-fail capacity check (avoids expensive session creation when full).
+        # Fast-fail capacity check (avoids expensive ChatSession creation when full).
        first_evicted: Workstream | None = None
        with self._lock:
            if len(self._workstreams) >= self._max_workstreams:
@@ -141,7 +142,7 @@ class WorkstreamManager:

            _m1.record_eviction()

-        # Create workstream and session outside the lock (session creation is
+        # Create workstream and ChatSession outside the lock (construction is
        # expensive — involves LLM client setup and DB writes).
        ws = Workstream(name=name)
        if ui_factory:
@@ -34,7 +34,6 @@ from turnstone.mq.protocol import (
    OutboundEvent,
    PlanReviewEvent,
    ReasoningEvent,
-    SessionResumedEvent,
    StateChangeEvent,
    StatusEvent,
    StreamEndEvent,
@@ -46,6 +45,7 @@ from turnstone.mq.protocol import (
    WorkstreamCreatedEvent,
    WorkstreamListEvent,
    WorkstreamRenameEvent,
+    WorkstreamResumedEvent,
 )

 if TYPE_CHECKING:
@@ -361,18 +361,21 @@ class Bridge:
        auto_approve_tools = getattr(msg, "auto_approve_tools", [])
        model = getattr(msg, "model", "")
        initial_message = getattr(msg, "initial_message", "")
-        resume_session = getattr(msg, "resume_session", "")
+        resume_ws = getattr(msg, "resume_ws", "")
+        user_id = getattr(msg, "user_id", "")
+        if user_id:
+            log.info("bridge.create_ws user_id=%s name=%s model=%s", user_id, name, model)
        ws_id, resumed = self._create_ws_on_server(
            name=name,
            auto_approve=auto_approve,
            auto_approve_tools=auto_approve_tools,
            correlation_id=msg.correlation_id,
            model=model,
-            resume_session=resume_session,
+            resume_ws=resume_ws,
        )
-        # Send initial_message only when no session was actually resumed.
+        # Send initial_message only when no workstream was actually resumed.
        # Use the server's `resumed` response (not just the intent) so that
-        # a pruned/missing session falls back to sending the initial message.
+        # a pruned/missing workstream falls back to sending the initial message.
        if ws_id and initial_message and not resumed:
            # Track the send so the global SSE handler emits TurnCompleteEvent
            # when the workstream returns to idle.
@@ -435,15 +438,15 @@ class Bridge:
        auto_approve_tools: list[str],
        correlation_id: str,
        model: str = "",
-        resume_session: str = "",
+        resume_ws: str = "",
    ) -> tuple[str, bool]:
        """Create a workstream on the server.  Returns (ws_id, resumed)."""
        try:
            payload: dict[str, Any] = {"name": name, "auto_approve": auto_approve}
            if model:
                payload["model"] = model
-            if resume_session:
-                payload["resume_session"] = resume_session
+            if resume_ws:
+                payload["resume_ws"] = resume_ws
            resp = self._http.post(
                "/v1/api/workstreams/new",
                json=payload,
@@ -472,15 +475,12 @@ class Bridge:

            self._start_ws_sse(ws_id)

-            resolved_session_id = data.get("session_id", "") if resumed else ""
-
            self._publish_global(
                WorkstreamCreatedEvent(
                    ws_id=ws_id,
                    name=ws_name,
                    correlation_id=correlation_id,
                    resumed=resumed,
-                    session_id=resolved_session_id,
                    message_count=data.get("message_count", 0),
                )
            )
@@ -497,10 +497,9 @@ class Bridge:
            if resumed:
                self._publish_ws(
                    ws_id,
-                    SessionResumedEvent(
+                    WorkstreamResumedEvent(
                        ws_id=ws_id,
                        correlation_id=correlation_id,
-                        session_id=resolved_session_id,
                        message_count=data.get("message_count", 0),
                        name=ws_name,
                    ),
@@ -95,7 +95,8 @@ class CreateWorkstreamMessage(InboundMessage):
    target_node: str = ""
    model: str = ""
    initial_message: str = ""
-    resume_session: str = ""
+    resume_ws: str = ""
+    user_id: str = ""


@dataclass
@@ -276,7 +277,6 @@ class WorkstreamCreatedEvent(OutboundEvent):
    name: str = ""
    node_id: str = ""
    resumed: bool = False
-    session_id: str = ""
    message_count: int = 0


@@ -336,11 +336,10 @@ class NodeListEvent(OutboundEvent):


@dataclass
-class SessionResumedEvent(OutboundEvent):
-    """Confirmation that a session was resumed during workstream creation."""
+class WorkstreamResumedEvent(OutboundEvent):
+    """Confirmation that a workstream was resumed during creation."""

-    type: str = "session_resumed"
-    session_id: str = ""
+    type: str = "ws_resumed"
    message_count: int = 0
    name: str = ""

@@ -410,7 +409,7 @@ _OUTBOUND_REGISTRY: dict[str, type[OutboundEvent]] = {
        ErrorEvent,
        InfoEvent,
        NodeListEvent,
-        SessionResumedEvent,
+        WorkstreamResumedEvent,
        ClusterStateEvent,
    ]
 }
@@ -25,12 +25,17 @@ from turnstone.api.schemas import (
    AuthLoginResponse,
    AuthSetupResponse,
    AuthStatusResponse,
+    ListScheduleRunsResponse,
+    ListSchedulesResponse,
+    ScheduleInfo,
    StatusResponse,
 )
 from turnstone.sdk._base import _BaseClient
 from turnstone.sdk._sync import _SyncRunner
 from turnstone.sdk.events import ClusterEvent

+_UNSET: Any = object()
+
 if TYPE_CHECKING:
    from collections.abc import AsyncIterator, Iterator

@@ -179,6 +184,109 @@ class AsyncTurnstoneConsole(_BaseClient):
    async def health(self) -> ConsoleHealthResponse:
        return await self._request("GET", "/health", response_model=ConsoleHealthResponse)

+    # -- schedules -----------------------------------------------------------
+
+    async def list_schedules(self) -> ListSchedulesResponse:
+        return await self._request(
+            "GET", "/v1/api/admin/schedules", response_model=ListSchedulesResponse
+        )
+
+    async def create_schedule(
+        self,
+        *,
+        name: str,
+        schedule_type: str,
+        initial_message: str,
+        description: str = "",
+        cron_expr: str = "",
+        at_time: str = "",
+        target_mode: str = "auto",
+        model: str = "",
+        auto_approve: bool = False,
+        auto_approve_tools: list[str] | None = None,
+        enabled: bool = True,
+    ) -> ScheduleInfo:
+        body: dict[str, Any] = {
+            "name": name,
+            "schedule_type": schedule_type,
+            "initial_message": initial_message,
+            "target_mode": target_mode,
+            "auto_approve": auto_approve,
+            "enabled": enabled,
+        }
+        if description:
+            body["description"] = description
+        if cron_expr:
+            body["cron_expr"] = cron_expr
+        if at_time:
+            body["at_time"] = at_time
+        if model:
+            body["model"] = model
+        if auto_approve_tools:
+            body["auto_approve_tools"] = auto_approve_tools
+        return await self._request(
+            "POST", "/v1/api/admin/schedules", json_body=body, response_model=ScheduleInfo
+        )
+
+    async def get_schedule(self, task_id: str) -> ScheduleInfo:
+        return await self._request(
+            "GET", f"/v1/api/admin/schedules/{task_id}", response_model=ScheduleInfo
+        )
+
+    async def update_schedule(
+        self,
+        task_id: str,
+        *,
+        name: Any = _UNSET,
+        description: Any = _UNSET,
+        schedule_type: Any = _UNSET,
+        cron_expr: Any = _UNSET,
+        at_time: Any = _UNSET,
+        target_mode: Any = _UNSET,
+        model: Any = _UNSET,
+        initial_message: Any = _UNSET,
+        auto_approve: Any = _UNSET,
+        auto_approve_tools: Any = _UNSET,
+        enabled: Any = _UNSET,
+    ) -> ScheduleInfo:
+        body: dict[str, Any] = {}
+        for key, val in [
+            ("name", name),
+            ("description", description),
+            ("schedule_type", schedule_type),
+            ("cron_expr", cron_expr),
+            ("at_time", at_time),
+            ("target_mode", target_mode),
+            ("model", model),
+            ("initial_message", initial_message),
+            ("auto_approve", auto_approve),
+            ("auto_approve_tools", auto_approve_tools),
+            ("enabled", enabled),
+        ]:
+            if val is not _UNSET:
+                body[key] = val
+        return await self._request(
+            "PUT",
+            f"/v1/api/admin/schedules/{task_id}",
+            json_body=body,
+            response_model=ScheduleInfo,
+        )
+
+    async def delete_schedule(self, task_id: str) -> StatusResponse:
+        return await self._request(
+            "DELETE", f"/v1/api/admin/schedules/{task_id}", response_model=StatusResponse
+        )
+
+    async def list_schedule_runs(
+        self, task_id: str, *, limit: int = 50
+    ) -> ListScheduleRunsResponse:
+        return await self._request(
+            "GET",
+            f"/v1/api/admin/schedules/{task_id}/runs",
+            params={"limit": limit},
+            response_model=ListScheduleRunsResponse,
+        )
+

 class TurnstoneConsole:
    """Synchronous client for the turnstone console API.
@@ -274,6 +382,84 @@ class TurnstoneConsole:
    def health(self) -> ConsoleHealthResponse:
        return self._runner.run(self._async.health())

+    # -- schedules -----------------------------------------------------------
+
+    def list_schedules(self) -> ListSchedulesResponse:
+        return self._runner.run(self._async.list_schedules())
+
+    def create_schedule(
+        self,
+        *,
+        name: str,
+        schedule_type: str,
+        initial_message: str,
+        description: str = "",
+        cron_expr: str = "",
+        at_time: str = "",
+        target_mode: str = "auto",
+        model: str = "",
+        auto_approve: bool = False,
+        auto_approve_tools: list[str] | None = None,
+        enabled: bool = True,
+    ) -> ScheduleInfo:
+        return self._runner.run(
+            self._async.create_schedule(
+                name=name,
+                schedule_type=schedule_type,
+                initial_message=initial_message,
+                description=description,
+                cron_expr=cron_expr,
+                at_time=at_time,
+                target_mode=target_mode,
+                model=model,
+                auto_approve=auto_approve,
+                auto_approve_tools=auto_approve_tools,
+                enabled=enabled,
+            )
+        )
+
+    def get_schedule(self, task_id: str) -> ScheduleInfo:
+        return self._runner.run(self._async.get_schedule(task_id))
+
+    def update_schedule(
+        self,
+        task_id: str,
+        *,
+        name: Any = _UNSET,
+        description: Any = _UNSET,
+        schedule_type: Any = _UNSET,
+        cron_expr: Any = _UNSET,
+        at_time: Any = _UNSET,
+        target_mode: Any = _UNSET,
+        model: Any = _UNSET,
+        initial_message: Any = _UNSET,
+        auto_approve: Any = _UNSET,
+        auto_approve_tools: Any = _UNSET,
+        enabled: Any = _UNSET,
+    ) -> ScheduleInfo:
+        return self._runner.run(
+            self._async.update_schedule(
+                task_id,
+                name=name,
+                description=description,
+                schedule_type=schedule_type,
+                cron_expr=cron_expr,
+                at_time=at_time,
+                target_mode=target_mode,
+                model=model,
+                initial_message=initial_message,
+                auto_approve=auto_approve,
+                auto_approve_tools=auto_approve_tools,
+                enabled=enabled,
+            )
+        )
+
+    def delete_schedule(self, task_id: str) -> StatusResponse:
+        return self._runner.run(self._async.delete_schedule(task_id))
+
+    def list_schedule_runs(self, task_id: str, *, limit: int = 50) -> ListScheduleRunsResponse:
+        return self._runner.run(self._async.list_schedule_runs(task_id, limit=limit))
+
    # -- lifecycle -----------------------------------------------------------

    def close(self) -> None:
@@ -26,7 +26,7 @@ from turnstone.api.server_schemas import (
    CreateWorkstreamResponse,
    DashboardResponse,
    HealthResponse,
-    ListSessionsResponse,
+    ListSavedWorkstreamsResponse,
    ListWorkstreamsResponse,
    SendResponse,
 )
@@ -76,7 +76,7 @@ class AsyncTurnstoneServer(_BaseClient):
        name: str = "",
        model: str = "",
        auto_approve: bool = False,
-        resume_session: str = "",
+        resume_ws: str = "",
    ) -> CreateWorkstreamResponse:
        body: dict[str, Any] = {}
        if name:
@@ -85,8 +85,8 @@ class AsyncTurnstoneServer(_BaseClient):
            body["model"] = model
        if auto_approve:
            body["auto_approve"] = True
-        if resume_session:
-            body["resume_session"] = resume_session
+        if resume_ws:
+            body["resume_ws"] = resume_ws
        return await self._request(
            "POST",
            "/v1/api/workstreams/new",
@@ -214,10 +214,12 @@ class AsyncTurnstoneServer(_BaseClient):
                    await consume_task
        return result

-    # -- sessions ------------------------------------------------------------
+    # -- saved workstreams ----------------------------------------------------

-    async def list_sessions(self) -> ListSessionsResponse:
-        return await self._request("GET", "/v1/api/sessions", response_model=ListSessionsResponse)
+    async def list_saved_workstreams(self) -> ListSavedWorkstreamsResponse:
+        return await self._request(
+            "GET", "/v1/api/workstreams/saved", response_model=ListSavedWorkstreamsResponse
+        )

    # -- auth ----------------------------------------------------------------

@@ -307,11 +309,11 @@ class TurnstoneServer:
        name: str = "",
        model: str = "",
        auto_approve: bool = False,
-        resume_session: str = "",
+        resume_ws: str = "",
    ) -> CreateWorkstreamResponse:
        return self._runner.run(
            self._async.create_workstream(
-                name=name, model=model, auto_approve=auto_approve, resume_session=resume_session
+                name=name, model=model, auto_approve=auto_approve, resume_ws=resume_ws
            )
        )

@@ -363,10 +365,10 @@ class TurnstoneServer:
            self._async.send_and_wait(message, ws_id, timeout=timeout, on_event=on_event)
        )

-    # -- sessions ------------------------------------------------------------
+    # -- saved workstreams ----------------------------------------------------

-    def list_sessions(self) -> ListSessionsResponse:
-        return self._runner.run(self._async.list_sessions())
+    def list_saved_workstreams(self) -> ListSavedWorkstreamsResponse:
+        return self._runner.run(self._async.list_saved_workstreams())

    # -- auth ----------------------------------------------------------------

@@ -302,7 +302,7 @@ class WebUI:
 def _build_history(
    session: ChatSession, has_pending_approval: bool = False
 ) -> list[dict[str, Any]]:
-    """Build a history replay list from session messages.
+    """Build a history replay list from ChatSession messages.

    When ``has_pending_approval`` is True, the last assistant entry's
    tool_calls are marked ``"pending": True`` so the client renders them
@@ -566,7 +566,6 @@ async def list_workstreams(request: Request) -> JSONResponse:
                "id": ws.id,
                "name": ws.name,
                "state": ws.state.value,
-                "session_id": ws.session.session_id if ws.session else None,
            }
        )
    return JSONResponse({"workstreams": result})
@@ -574,7 +573,7 @@ async def list_workstreams(request: Request) -> JSONResponse:

 async def dashboard(request: Request) -> JSONResponse:
    """GET /v1/api/dashboard — enriched workstream data + aggregate stats."""
-    from turnstone.core.memory import get_session_name
+    from turnstone.core.memory import get_workstream_display_name

    mgr: WorkstreamManager = request.app.state.workstreams
    wss = mgr.list_all()
@@ -596,13 +595,12 @@ async def dashboard(request: Request) -> JSONResponse:
            active_count += 1
        title = ""
        if ws.session:
-            title = get_session_name(ws.session.session_id) or ""
+            title = get_workstream_display_name(ws.session.ws_id) or ""
        ws_list.append(
            {
                "id": ws.id,
                "name": ws.name,
                "state": ws.state.value,
-                "session_id": ws.session.session_id if ws.session else None,
                "title": title,
                "tokens": tok,
                "context_ratio": round(ctx, 3),
@@ -630,25 +628,23 @@ async def dashboard(request: Request) -> JSONResponse:
    )


-async def list_sessions_endpoint(request: Request) -> JSONResponse:
-    """GET /v1/api/sessions — list saved sessions."""
-    from turnstone.core.memory import list_sessions
+async def list_saved_workstreams(request: Request) -> JSONResponse:
+    """GET /v1/api/workstreams/saved — list saved workstreams with conversation history."""
+    from turnstone.core.memory import list_workstreams_with_history

-    rows = list_sessions(limit=50)
-    sessions = [
+    rows = list_workstreams_with_history(limit=50)
+    result = [
        {
-            "session_id": sid,
+            "ws_id": wid,
            "alias": alias,
            "title": title,
            "created": created,
            "updated": updated,
            "message_count": count,
-            "node_id": node_id,
-            "ws_id": ws_id,
        }
-        for sid, alias, title, created, updated, count, node_id, ws_id in rows
+        for wid, alias, title, created, updated, count, *_extra in rows
    ]
-    return JSONResponse({"sessions": sessions})
+    return JSONResponse({"workstreams": result})


 def _count_ws_states(wss: list[Workstream]) -> dict[str, int]:
@@ -694,7 +690,6 @@ async def metrics_endpoint(request: Request) -> Response:
                {
                    "ws_id": ws.id,
                    "name": ws.name,
-                    "session_id": ws.session.session_id if ws.session else "",
                    "prompt_tokens": ui._ws_prompt_tokens,
                    "completion_tokens": ui._ws_completion_tokens,
                    "messages": ui._ws_messages,
@@ -814,7 +809,7 @@ async def command(request: Request) -> JSONResponse:
        should_exit = ws.session.handle_command(cmd)
        if should_exit:
            ui.on_info("Session ended. You can close this tab.")
-        # Handle UI updates for session-changing commands
+        # Handle UI updates for workstream-changing commands
        cmd_word = cmd.strip().split(None, 1)[0].lower()
        if cmd_word in ("/clear", "/new"):
            ui._enqueue({"type": "clear_ui"})
@@ -826,9 +821,9 @@ async def command(request: Request) -> JSONResponse:
        # Sync in-memory workstream name after any command that can change it.
        # This ensures /api/workstreams and future page loads see the right name.
        if cmd_word in ("/name", "/resume"):
-            from turnstone.core.memory import get_session_name
+            from turnstone.core.memory import get_workstream_display_name

-            updated_name = get_session_name(ws.session.session_id)
+            updated_name = get_workstream_display_name(ws.session.ws_id) if ws.session else None
            if updated_name:
                ws.name = updated_name
    except Exception as e:
@@ -868,20 +863,18 @@ async def create_workstream(request: Request) -> JSONResponse:
                        "reason": "evicted",
                    }
                )
-        # Atomic session resume during creation.
+        # Atomic workstream resume during creation.
        resumed = False
        message_count = 0
-        session_id = ""
-        resume_session_id = body.get("resume_session", "")
-        if resume_session_id and ws.session is not None:
-            from turnstone.core.memory import get_session_name, resolve_session
+        resume_ws_id = body.get("resume_ws", "")
+        if resume_ws_id and ws.session is not None:
+            from turnstone.core.memory import get_workstream_display_name, resolve_workstream

-            target_id = resolve_session(resume_session_id)
-            if target_id and ws.session.resume_session(target_id):
+            target_id = resolve_workstream(resume_ws_id)
+            if target_id and ws.session.resume(target_id):
                resumed = True
-                session_id = target_id
                message_count = len(ws.session.messages)
-                ws.name = get_session_name(target_id) or ws.name
+                ws.name = get_workstream_display_name(target_id) or ws.name
                ui = ws.ui
                if isinstance(ui, WebUI):
                    ui._enqueue({"type": "clear_ui"})
@@ -894,7 +887,6 @@ async def create_workstream(request: Request) -> JSONResponse:
                "ws_id": ws.id,
                "name": ws.name,
                "resumed": resumed,
-                "session_id": session_id,
                "message_count": message_count,
            }
        )
@@ -1084,7 +1076,7 @@ def create_app(
                    Route("/api/events/global", global_events_sse),
                    Route("/api/workstreams", list_workstreams),
                    Route("/api/dashboard", dashboard),
-                    Route("/api/sessions", list_sessions_endpoint),
+                    Route("/api/workstreams/saved", list_saved_workstreams),
                    Route("/api/send", send_message, methods=["POST"]),
                    Route("/api/approve", approve, methods=["POST"]),
                    Route("/api/plan", plan_feedback, methods=["POST"]),
@@ -1223,8 +1215,8 @@ def main() -> None:
    parser.add_argument(
        "--resume",
        default=None,
-        metavar="SESSION",
-        help="Resume a previous session by alias or session_id",
+        metavar="WS",
+        help="Resume a previous workstream by alias or ws_id",
    )
    parser.add_argument(
        "--skip-permissions",
@@ -1248,11 +1240,11 @@ def main() -> None:
        help="Port to listen on (default: 8080)",
    )
    parser.add_argument(
-        "--session-retention-days",
+        "--retention-days",
        type=int,
        default=90,
        metavar="DAYS",
-        help="Delete unnamed sessions older than DAYS days on startup, 0 to disable (default: 90)",
+        help="Delete unnamed workstreams older than DAYS days on startup, 0 to disable (default: 90)",
    )
    parser.add_argument(
        "--workstream-idle-timeout",
@@ -1348,10 +1340,10 @@ def main() -> None:
    )
    init_storage(db_backend, path=db_path, url=db_url, pool_size=db_pool_size)

-    # Prune stale / empty sessions on startup
-    from turnstone.core.memory import prune_sessions
+    # Prune stale / empty workstreams on startup
+    from turnstone.core.memory import prune_workstreams

-    prune_sessions(retention_days=args.session_retention_days, log_fn=print)
+    prune_workstreams(retention_days=args.retention_days, log_fn=print)

    # Create client and detect model
    provider_name = args.provider
@@ -1488,16 +1480,16 @@ def main() -> None:
    # Handle --resume
    assert ws.session is not None
    if args.resume:
-        from turnstone.core.memory import resolve_session
+        from turnstone.core.memory import resolve_workstream

-        target_id = resolve_session(args.resume)
+        target_id = resolve_workstream(args.resume)
        if not target_id:
-            log.error("Session not found: %s", args.resume)
+            log.error("Workstream not found: %s", args.resume)
            sys.exit(1)
-        if not ws.session.resume_session(target_id):
-            log.error("Session '%s' has no messages.", args.resume)
+        if not ws.session.resume(target_id):
+            log.error("Workstream '%s' has no messages.", args.resume)
            sys.exit(1)
-        log.info("Resumed session %s (%d messages)", target_id, len(ws.session.messages))
+        log.info("Resumed workstream %s (%d messages)", target_id, len(ws.session.messages))

    # Record detected model in metrics
    _metrics.model = model
@@ -0,0 +1,35 @@
+{
+  "name": "notify",
+  "description": "Send a notification to a user or channel on an external platform (Discord, etc.). Use this to proactively alert people about task completion, errors, or important updates. Provide either 'username' for user-based targeting (sends to all linked channels) or 'channel_type' + 'channel_id' for direct targeting. Do not combine both.",
+  "parameters": {
+    "type": "object",
+    "properties": {
+      "message": {
+        "type": "string",
+        "description": "The notification content (plain text, max 2000 chars)."
+      },
+      "username": {
+        "type": "string",
+        "description": "Turnstone username to notify. Sends to all linked channels for this user."
+      },
+      "channel_type": {
+        "type": "string",
+        "description": "Channel platform for direct targeting.",
+        "enum": ["discord"]
+      },
+      "channel_id": {
+        "type": "string",
+        "description": "Platform-specific channel or user ID for direct targeting."
+      },
+      "title": {
+        "type": "string",
+        "description": "Optional short title for the notification."
+      }
+    },
+    "required": ["message"]
+  },
+  "agent": true,
+  "task_agent": true,
+  "auto_approve": true,
+  "primary_key": "message"
+}
@@ -1,6 +1,6 @@
 {
  "name": "plan",
-  "description": "Plan before implementing. An autonomous agent explores the codebase and writes a structured plan to .plan-<session_id>.md (unique per session to avoid workstream collisions). If a plan for this session already exists it is re-read and refined rather than overwritten from scratch. Use plan BEFORE writing code — when the user asks to build, add, refactor, or change something that touches multiple files or has unclear scope. The plan identifies files to modify, existing patterns to reuse, and risks to consider.",
+  "description": "Plan before implementing. An autonomous agent explores the codebase and writes a structured plan to .plan-<ws_id>.md (unique per workstream to avoid collisions). If a plan for this workstream already exists it is re-read and refined rather than overwritten from scratch. Use plan BEFORE writing code — when the user asks to build, add, refactor, or change something that touches multiple files or has unclear scope. The plan identifies files to modify, existing patterns to reuse, and risks to consider.",
  "parameters": {
    "type": "object",
    "properties": {
@@ -617,12 +617,12 @@ function toggleDashboard() {
 function loadDashboard() {
  var tableEl = document.getElementById("dash-ws-table");
  tableEl.innerHTML = '<div class="dashboard-empty">Loading\u2026</div>';
-  document.getElementById("dashboard-session-cards").innerHTML =
+  document.getElementById("dashboard-saved-cards").innerHTML =
    '<div class="dashboard-empty">Loading\u2026</div>';
  var dashP = authFetch("/v1/api/dashboard").then(function (r) {
    return r.json();
  });
-  var sessP = authFetch("/v1/api/sessions").then(function (r) {
+  var sessP = authFetch("/v1/api/workstreams/saved").then(function (r) {
    return r.json();
  });
  Promise.all([dashP, sessP])
@@ -631,19 +631,19 @@ function loadDashboard() {
      var wsList = dashData.workstreams || [];
      var agg = dashData.aggregate || {};
      renderDashboardTable(wsList, agg);
-      // Collect active session IDs for dedup
-      var activeSessionIds = {};
+      // Collect active ws IDs for dedup
+      var activeWsIds = {};
      wsList.forEach(function (ws) {
-        if (ws.session_id) activeSessionIds[ws.session_id] = true;
+        activeWsIds[ws.id] = true;
      });
-      var sessList = (res[1].sessions || []).filter(function (s) {
-        return !activeSessionIds[s.session_id];
+      var savedList = (res[1].workstreams || []).filter(function (s) {
+        return !activeWsIds[s.ws_id];
      });
-      renderDashboardSessions(sessList);
+      renderSavedWorkstreams(savedList);
    })
    .catch(function () {
      tableEl.innerHTML = '<div class="dashboard-empty">Failed to load</div>';
-      document.getElementById("dashboard-session-cards").innerHTML =
+      document.getElementById("dashboard-saved-cards").innerHTML =
        '<div class="dashboard-empty">Failed to load</div>';
    });
 }
@@ -800,30 +800,30 @@ function updateDashFooter(agg) {
      ")";
  }
 }
-function renderDashboardSessions(sessions) {
-  var c = document.getElementById("dashboard-session-cards");
+function renderSavedWorkstreams(items) {
+  var c = document.getElementById("dashboard-saved-cards");
  c.innerHTML = "";
-  if (!sessions.length) {
-    c.innerHTML = '<div class="dashboard-empty">No saved sessions</div>';
+  if (!items.length) {
+    c.innerHTML = '<div class="dashboard-empty">No saved workstreams</div>';
    return;
  }
-  sessions.forEach(function (sess) {
+  items.forEach(function (sess) {
    var card = document.createElement("div");
    card.className = "dashboard-card";
    card.setAttribute("role", "button");
    card.setAttribute("tabindex", "0");
-    var label = sess.alias || sess.title || sess.session_id;
+    var label = sess.alias || sess.title || sess.ws_id;
    card.setAttribute("aria-label", "Resume: " + label);
    card.onclick = function () {
-      dashboardResumeSession(sess.session_id);
+      dashboardResumeSession(sess.ws_id);
    };
    card.onkeydown = function (e) {
      if (e.key === "Enter" || e.key === " ") {
        e.preventDefault();
-        dashboardResumeSession(sess.session_id);
+        dashboardResumeSession(sess.ws_id);
      }
    };
-    var title = sess.alias || sess.title || sess.session_id.substring(0, 12);
+    var title = sess.alias || sess.title || sess.ws_id.substring(0, 12);
    var meta = sess.message_count + " msgs";
    if (sess.updated) meta += " \u00b7 " + formatRelativeTime(sess.updated);
    card.innerHTML =
@@ -861,11 +861,11 @@ function dashboardSwitchWorkstream(wsId) {
    switchTab(wsId);
  } else loadDashboard();
 }
-function dashboardResumeSession(sessionId) {
+function dashboardResumeSession(wsId) {
  authFetch("/v1/api/workstreams/new", {
    method: "POST",
    headers: { "Content-Type": "application/json" },
-    body: JSON.stringify({ resume_session: sessionId }),
+    body: JSON.stringify({ resume_ws: wsId }),
  })
    .then(function (r) {
      if (!r.ok) throw new Error("HTTP " + r.status);
@@ -879,7 +879,7 @@ function dashboardResumeSession(sessionId) {
      // Resume handled atomically by server — history arrives via SSE.
    })
    .catch(function (err) {
-      showToast("Failed to resume session", "error");
+      showToast("Failed to resume workstream", "error");
    });
 }
 function dashboardNewChat() {
@@ -62,9 +62,9 @@
      <span class="dash-footer-nodes" id="dash-footer-nodes"></span>
      <span class="dash-footer-stats" id="dash-footer-stats"></span>
    </div>
-    <section class="dashboard-section" id="dashboard-sessions" aria-label="Recent sessions">
-      <h2 class="dashboard-section-title">Recent Sessions</h2>
-      <div class="dashboard-cards" id="dashboard-session-cards"></div>
+    <section class="dashboard-section" id="dashboard-saved-ws" aria-label="Saved workstreams">
+      <h2 class="dashboard-section-title">Saved Workstreams</h2>
+      <div class="dashboard-cards" id="dashboard-saved-cards"></div>
    </section>
  </div>
 </div>
Author	SHA1	Message	Date
Patrick Buckley	8b92302247	Bump version to 0.4.3	2026-03-07 13:11:42 -08:00
Patrick Buckley	e195ca54a6	new high level arch abstract	2026-03-07 13:01:19 -08:00
Patrick Buckley	50277cd4de	Add 10-node cluster profile to Docker Compose	2026-03-07 12:54:41 -08:00
Patrick Buckley	fb190f8977	Normalize session_id into ws_id as sole persistent identity (#29 ) * Normalize session_id into ws_id as sole persistent identity Eliminate the separate session_id concept. The workstream ID (ws_id) is now the single identity used for both real-time routing and conversation persistence, removing a layer of indirection that was 1:1 in practice and buggy on resume (stale pointers, orphaned rows). Schema changes (migration 006): - Drop sessions table; add alias/title columns to workstreams - Rename conversations.session_id → ws_id - Rename session_config table → workstream_config (ws_id column) - Data migration remaps existing conversations to ws_id Storage/API renames: - register_session → register_workstream (already existed, merged) - save_message/load_messages now keyed by ws_id - resolve_session → resolve_workstream - ChatSession.session_id property → ws_id - ChatSession.resume_session() → resume() - resume_session field → resume_ws - SessionResumedEvent → WorkstreamResumedEvent - /api/sessions → /api/workstreams/saved - /sessions slash command → /workstreams - --session-retention-days → --retention-days Channel eviction recovery simplified: reuses old ws_id directly instead of get_session_id_by_ws() reverse lookup. * Fix Copilot review feedback: stale session wording in docs, regenerate OpenAPI spec - docs/channels.md: "resumes the session" → "resumes the workstream", "Session resumed:" → "Resumed:", "old session was pruned" → "old workstream was pruned" - docs/api-reference.md: "Each session object" → "Each saved workstream object", field descriptions updated, removed stale node_id field - sdk/typescript/openapi-server.json: fully regenerated from Python models — removes all stale session_id properties from WorkstreamInfo, DashboardWorkstream, CreateWorkstreamResponse schemas	2026-03-07 12:49:07 -08:00
Patrick Buckley	25b5e32089	Bump version to 0.4.2	2026-03-05 20:18:46 -08:00
Patrick Buckley	339981a258	Add GPT-5.3, GPT-5.4, and pro model capabilities (#28 ) * Add GPT-5.3, GPT-5.4, and pro model capabilities Add capability entries for gpt-5-pro (272k output, high-only reasoning), gpt-5.2-pro, gpt-5.3, gpt-5.4 (1.05M context), and gpt-5.4-pro. * Validate reasoning_effort against model capabilities _apply_model_params now falls back to caps.default_reasoning_effort when the requested value is not in caps.reasoning_effort_values. Prevents sending unsupported effort levels to models like gpt-5-pro (high only).	2026-03-05 20:17:03 -08:00
Patrick Buckley	06de9ff83b	Bump version to 0.4.1	2026-03-05 18:13:42 -08:00
Patrick Buckley	924b976f1f	Add scheduled task docs and SDK client methods Add documentation and SDK support for the scheduled task system (cron/at scheduling via console API). Includes Python SDK methods (async + sync), TypeScript SDK methods, console.md API reference, sdk.md table update, and architecture.md module map entry.	2026-03-05 18:11:36 -08:00
Patrick Buckley	d5db817391	Fix channel gateway Docker networking and console proxy approval scope Two runtime bugs: 1. Channel gateway advertised http://127.0.0.1:8091 which is unreachable from other Docker containers. Add TURNSTONE_CHANNEL_ADVERTISE_URL env var override for Docker/K8s environments, set to http://channel:8091 in compose.yaml, and pass --http-host=0.0.0.0 so the gateway listens on all interfaces. 2. Console proxy service JWT had only "write" scope but the approval endpoint requires "approve". Tool approval buttons in the server web UI silently failed when accessed through the console proxy. Changed proxy token scopes to read+write+approve.	2026-03-05 18:01:52 -08:00
Patrick Buckley	fc8ceb4c72	Update README: 3-node cluster diagram, remove directory tree Replace single-node Mermaid diagram with a 3-node cluster layout showing bridge+server pairs per node, shared Redis MQ, console, and channel gateway. Remove the verbose directory tree listing.	2026-03-05 17:44:33 -08:00
Patrick Buckley	07234dec4d	Replace ASCII architecture diagram with Mermaid in README GitHub renders Mermaid natively as an interactive SVG. The new diagram shows all client entry points (CLI, browser, SDK, Discord), the full cluster topology including the channel gateway and notify path, and the LLM provider layer.	2026-03-05 17:34:42 -08:00
Patrick Buckley	dd4cc0b30d	Add PROGRESS.md and .coverage to .gitignore	2026-03-05 17:31:55 -08:00
Patrick Buckley	e7fe8fca9d	Add channel notification tool with security hardening (#27 ) * Add channel notification tool with security hardening Implements the `notify` tool allowing the LLM to send notifications to Discord channels/users via the channel gateway. Includes fixes for 11 review findings: JWT auth on the gateway endpoint, first-healthy gateway delivery with retry+backoff, rate limiting only on success, SSRF URL scheme validation, Discord mention sanitization, SQLite ON CONFLICT upsert preserving created timestamps, advertise URL resolution for 0.0.0.0 bind, randomized service IDs, generic error messages to prevent internal state leakage, and partial direct-target validation. Service registry with heartbeat-based health filtering (migration 005). Channel gateway registers on startup, heartbeats every 30s, deregisters on shutdown. 70 new tests covering tool prepare/execute, HTTP endpoint auth (static + JWT), storage CRUD, and retry behavior. * Add notify documentation, diagrams, and review fixes Documentation: - New sequence diagram 17-notify-flow.puml showing end-to-end delivery - Updated 16-channel-architecture.puml with services table, notify HTTP path, and Notification Flow note - channels.md: Notifications section (targeting, delivery flow, service registry, security) and new config table entries - tools.md: notify tool reference, updated counts/tables (14→15 tools) - security.md: channel gateway row in service-to-service auth table - architecture.md: notification subsystem paragraph Review fixes (copilot): - _http.py: fail closed when auth unconfigured (401 instead of pass- through), strip whitespace on message/title, generic error messages for user-not-found vs no-linked-channels - session.py: parse gateway response JSON and require at least one result with status=="sent" before counting as success - _postgresql.py: use index_elements instead of constraint for upsert	2026-03-05 17:24:00 -08:00
Patrick Buckley	42b9f89988	Add scheduled task system with cron/at scheduling (#26 ) * Add scheduled task system with cron/at scheduling, admin API, and console UI Console-integrated background scheduler dispatches workstreams on recurring cron expressions or one-shot ISO8601 timestamps. Four target modes: auto (best node by headroom), pool (shared queue), all (fan-out), or specific node. Redis distributed lock with unique owner + Lua conditional release prevents duplicate dispatch in multi-console deployments. Storage: scheduled_tasks + scheduled_task_runs tables (migration 004), 9 protocol methods on both SQLite and PostgreSQL backends, field allowlist on updates, run history auto-pruned at 90 days. API: 6 CRUD endpoints under /v1/api/admin/schedules with croniter validation, ISO8601 future-time checks, field length bounds, schedule count cap (200), and OpenAPI spec entries with Pydantic models. UI: Schedules tab in admin panel with create/edit/delete modals, run history modal, cron/at type toggle, target mode select, status dots for accessibility, responsive grid, keyboard navigation, and focus management. Security: fan-out capped at 20 nodes/task/tick, auto-approve dispatches logged at WARNING with created_by attribution, user_id propagated in CreateWorkstreamMessage for audit trail. 46 tests across storage, scheduler engine, and API endpoints. * Add croniter to test extras for CI compatibility CI installs [test] extras but not [console], so croniter was missing when schedule API tests import console/server.py validation functions. * Address Copilot review: timezone validation, focus trap, enabled flag - Reject naive at_time timestamps — require timezone offset (e.g. +00:00 or Z) - UI appends +00:00 to datetime-local values for explicit UTC - Fix datetime-local normalization: check length before appending seconds - Add textarea to modal focus trap selector (prevents focus escape) - Fix _normalize_task_dict not called in update response - Persist enabled=false on create (storage defaults to enabled=1) - Validate at_time is still in future when re-enabling a one-shot task - broker._redis coupling acknowledged as tracked tech debt	2026-03-05 16:05:00 -08:00