bump: v0.8.9 — mTLS + ACME integration via lacme

* fix: TLS Docker end-to-end testing fixes

Fixes discovered during Docker Compose TLS integration testing:

- Dockerfile: use --extra all (prevents missing optional deps)
- lacme 1.0.3: fixes CACertificateIssued event logging crash
- chmod PermissionError: guard for Docker volume mounts
- socket import: moved to top of main() (was inside TLS conditional,
  caused NameError in _default_node_id)
- redis.SSLConnection: ConnectionPool needs explicit connection_class,
  not ssl=True (which only works on Redis() directly)
- Empty redis password: pass None instead of "" to avoid AUTH error
- TURNSTONE_CONSOLE_URL: env var for Docker service discovery
  (0.0.0.0 bind address isn't reachable from other containers)
- HTTP01Handler: ACME client needs a challenge handler even when
  server auto-approves
- Docker overlay: tls-init as root with chmod, Redis conditional
  password, console Redis TLS flags, TURNSTONE_CONSOLE_URL

* feat: full mTLS end-to-end with lacme 1.0.4

Completes the mTLS chain across all services:

lacme 1.0.4:
- Dual EKU certs (serverAuth + clientAuth) — fixes mTLS rejection
- Configurable CA name (name="turnstone") — consistent store key

Bootstrap CA import:
- Console imports bootstrap CA from /certs volume on first boot
- Single trust root: bootstrap CA → console → all service certs

Bridge mTLS:
- TLSClient init when TURNSTONE_TLS_ENABLED set
- Auto-upgrades server URL from http:// to https://
- SSLContext passed to all 3 httpx clients via verify=

Console collector mTLS:
- upgrade_tls() method replaces httpx client with mTLS context
- Called in lifespan after cert issuance alongside proxy upgrade
- Fixes "Failed to poll node" when server serves HTTPS

Docker overlay:
- TURNSTONE_TLS_SANS on all services (Docker service names as SANs)
- TURNSTONE_TLS_ENABLED on bridge
- Channel service with Redis TLS flags
- TURNSTONE_CONSOLE_URL for service discovery
- Server healthcheck disabled (mTLS healthcheck deferred)
- Redis conditional password from env

Verified end-to-end: bootstrap → console CA → server HTTPS →
bridge mTLS → Redis TLS → channel Redis TLS → console collector
polls server over mTLS → workstream creation works through bridge

* fix: lint + copilot feedback on TLS Docker e2e

- SIM105: contextlib.suppress(PermissionError) for chmod
- F401: remove unused get_storage import in bridge
- Redis healthcheck: pass password when REDIS_PASSWORD is set

* fix: sort imports in admin.py and bridge.py

* fix: tls-init key permissions, healthcheck env, collector race

- tls-init: add set -e, chown to turnstone:turnstone with restrictive
  perms (keys 0600, certs 0640, dirs 0750) instead of world-readable
- Redis healthcheck: use container runtime $$REDIS_PASSWORD instead of
  Compose-time interpolation for consistency with --requirepass block
- collector upgrade_tls(): don't close old httpx client while concurrent
  poll threads may still be using it — let GC handle cleanup

.github/workflows/ci.yml

@@ -47,11 +47,37 @@ jobs:
           name: coverage-${{ matrix.python-version }}
           path: coverage.xml
 
+  test-postgres:
+    runs-on: ubuntu-latest
+    services:
+      postgres:
+        image: postgres:18
+        env:
+          POSTGRES_USER: postgres
+          POSTGRES_PASSWORD: postgres
+          POSTGRES_DB: turnstone_test
+        ports:
+          - 5432:5432
+        options: >-
+          --health-cmd="pg_isready -U postgres"
+          --health-interval=10s
+          --health-timeout=5s
+          --health-retries=5
+    steps:
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6
+      - uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6
+        with:
+          python-version: "3.12"
+      - run: pip install -e ".[test,mq,postgres]"
+      - run: pytest tests/ -m "not live" --storage-backend=postgresql -q
+        env:
+          TURNSTONE_TEST_PG_URL: postgresql+psycopg://postgres:postgres@localhost:5432/turnstone_test
+
   lock-check:
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6
-      - uses: astral-sh/setup-uv@e06108dd0aef18192324c70427afc47652e63a82 # v7
+      - uses: astral-sh/setup-uv@37802adc94f370d6bfd71619e3f0bf239e1f3b78 # v7
         with:
           uv-version: "0.9.18"
       - run: uv lock --check
@@ -60,7 +86,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6
-      - uses: astral-sh/setup-uv@e06108dd0aef18192324c70427afc47652e63a82 # v7
+      - uses: astral-sh/setup-uv@37802adc94f370d6bfd71619e3f0bf239e1f3b78 # v7
         with:
           uv-version: "0.9.18"
       - uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6

Dockerfile

@@ -8,7 +8,7 @@ FROM python:3.14-slim
 LABEL org.opencontainers.image.title="turnstone" \
       org.opencontainers.image.description="Multi-node AI orchestration platform"
 
-COPY --from=ghcr.io/astral-sh/uv:0.10.10 /uv /usr/local/bin/uv
+COPY --from=ghcr.io/astral-sh/uv:0.11.1 /uv /usr/local/bin/uv
 
 # System dependencies for psycopg (PostgreSQL client library)
 RUN apt-get update && apt-get upgrade -y && apt-get install -y --no-install-recommends libpq5 \
@@ -25,12 +25,12 @@ ENV UV_COMPILE_BYTECODE=1
 # Install dependencies first (cached layer — only re-runs when deps change)
 COPY pyproject.toml uv.lock README.md LICENSE ./
 RUN uv sync --frozen --no-install-project --no-dev \
-    --extra mq --extra console --extra sim --extra postgres --extra discord --extra anthropic
+    --extra all
 
 # Install the project itself
 COPY turnstone/ turnstone/
 RUN uv sync --frozen --no-dev \
-    --extra mq --extra console --extra sim --extra postgres --extra discord --extra anthropic
+    --extra all
 
 # Add venv to PATH so entry points are found
 ENV PATH="/app/.venv/bin:$PATH"

README.md

@@ -5,9 +5,11 @@
 [![Python](https://img.shields.io/pypi/pyversions/turnstone)](https://pypi.org/project/turnstone/)
 [![License](https://img.shields.io/badge/license-BSL--1.1-blue)](LICENSE)
 
-Multi-node AI orchestration platform. Deploy tool-using AI agents across a cluster of servers, driven by message queues or interactive interfaces.
+Experimental multi-node AI orchestration platform. Deploy tool-using AI agents across a cluster of servers, driven by message queues or interactive interfaces.
 
-Named after the [Ruddy Turnstone](https://en.wikipedia.org/wiki/Ruddy_turnstone) — a bird that flips rocks to expose what's hiding underneath.
+> **Beta — Use at your own risk.** Turnstone is under active development and has not reached a stable release. APIs, configuration formats, and database schemas may change between versions without migration paths. We make no guarantees of determinism, reliability, or backward compatibility. Evaluate thoroughly before deploying to any environment where these properties matter.
+
+Named after the [Ruddy Turnstone](https://en.wikipedia.org/wiki/Ruddy_turnstone) (*Arenaria interpres*) — a shorebird that flips stones to discover what's hiding underneath.
 
 ## What it does
 
@@ -145,7 +147,7 @@ Turnstone includes a built-in governance layer for enterprise deployments — ma
 - **RBAC** — 15 granular permissions, 3 built-in roles (admin / operator / viewer), custom roles, privilege escalation prevention
 - **OIDC SSO** — single sign-on via any OpenID Connect provider (Okta, Azure AD, Google, Keycloak); Authorization Code Flow with PKCE, auto-provisioning, claim-based role mapping with demotion propagation; see [docs/oidc.md](docs/oidc.md)
 - **Tool policies** — glob-pattern rules (`allow` / `deny` / `ask`) with priority ordering; automate approvals or lock down dangerous tools
-- **Skills** — reusable behavioral profiles with system prompts, `{{variable}}` substitution, session config (model, temperature, token budget), install-time security scanning, version history, external discovery (skills.sh / GitHub), and runtime `load_skill` tool for model-driven skill activation
+- **Skills** — reusable behavioral profiles with system prompts, `{{variable}}` substitution, session config (model, temperature, token budget), install-time security scanning, version history, external discovery (skills.sh / GitHub), and runtime `skill` tool for model-driven skill activation
 - **Usage tracking** — per-request token and tool metrics, aggregation by day / model / user, automatic 90-day pruning
 - **Audit logging** — append-only event trail for all admin mutations, IP-aware, 365-day retention
 
@@ -308,7 +310,7 @@ search_max_results = 5     # max tools returned per search query
 [server]
 host = "0.0.0.0"
 port = 8080
-max_workstreams = 10       # auto-evicts oldest idle when full
+max_workstreams = 50       # auto-evicts oldest idle when full
 
 [redis]
 host = "localhost"
@@ -340,7 +342,7 @@ burst = 20
 backend = "sqlite"     # "sqlite" (default) or "postgresql"
 path = ".turnstone.db" # SQLite file path (relative to working directory)
 # url = "postgresql+psycopg://user:pass@host:5432/turnstone"  # PostgreSQL
-# pool_size = 5        # PostgreSQL connection pool size
+# pool_size = 2        # PostgreSQL connection pool size (per process)
 
 [judge]
 enabled = true         # intent validation for tool approvals (--no-judge to disable)
@@ -394,7 +396,7 @@ Idle workstreams are automatically cleaned up after 2 hours (configurable). In m
 - `turnstone_judge_llm_latency_seconds` — LLM judge evaluation latency histogram
 - `turnstone_judge_enabled` — whether the intent validation judge is active (0/1)
 
-Per-workstream metrics are labeled by `ws_id` (bounded to 10 max workstreams).
+Per-workstream metrics are labeled by `ws_id` (bounded by `[server].max_workstreams`).
 
 ### Health & Rate Limiting
 
@@ -404,7 +406,7 @@ Per-workstream metrics are labeled by `ws_id` (bounded to 10 max workstreams).
 
 **Per-IP rate limiting.** When `[ratelimit].enabled` is true, each client IP is tracked with a token-bucket limiter (`requests_per_second` / `burst`). Rate limiting is applied in `do_GET`/`do_POST` after authentication but before route dispatch. `/health` and `/metrics` are exempt. Requests that exceed the limit receive HTTP 429 with a `Retry-After` header.
 
-**Workstream eviction.** When `WorkstreamManager.create()` would exceed `max_workstreams`, the oldest IDLE workstream is automatically evicted and the `turnstone_workstreams_evicted_total` counter is incremented. Configure via `[server].max_workstreams` (default 10).
+**Workstream eviction.** When `WorkstreamManager.create()` would exceed `max_workstreams`, the oldest IDLE workstream is automatically evicted and the `turnstone_workstreams_evicted_total` counter is incremented. Configure via `[server].max_workstreams` (default 50).
 
 ## Requirements
 

deploy/docker-compose.tls.yml

@@ -0,0 +1,146 @@
+# TLS overlay — enables mTLS across the turnstone cluster.
+#
+# Usage (requires base compose.yaml with production profile):
+#   docker compose -f compose.yaml -f deploy/docker-compose.tls.yml --profile production up
+#
+# The tls-init service bootstraps a CA and issues a cert for Redis.
+# All turnstone services auto-provision their own certs via the
+# console's ACME endpoint.
+
+services:
+  # Bootstrap: create CA + Redis cert before anything starts.
+  # Runs as root to create directories in the volume, then chowns
+  # to turnstone:turnstone with restrictive perms (keys 0600).
+  tls-init:
+    build: .
+    user: root
+    command:
+      - sh
+      - -c
+      - |
+        set -e
+        turnstone-admin tls-bootstrap --out /certs --issue redis
+        chown -R turnstone:turnstone /certs
+        find /certs -type d -exec chmod 750 {} +
+        find /certs -type f -name '*key.pem' -exec chmod 600 {} +
+        find /certs -type f ! -name '*key.pem' -exec chmod 640 {} +
+    volumes:
+      - tls-certs:/certs
+    networks:
+      - turnstone-net
+    restart: "no"
+
+  # Console: runs the internal CA + ACME server
+  console:
+    depends_on:
+      tls-init:
+        condition: service_completed_successfully
+    volumes:
+      - tls-certs:/certs:ro
+    environment:
+      TURNSTONE_TLS_ENABLED: "true"
+      TURNSTONE_TLS_SANS: "console"
+      TURNSTONE_CONSOLE_URL: "http://console:8090"
+    command:
+      - turnstone-console
+      - --host=0.0.0.0
+      - --port=8090
+      - --redis-host=redis
+      - --redis-port=6379
+      - --poll-interval=${CONSOLE_POLL_INTERVAL:-10}
+      - --redis-tls
+      - --redis-tls-ca=/certs/ca.pem
+
+  # Server: auto-provisions certs via console ACME, serves HTTPS
+  server:
+    depends_on:
+      console:
+        condition: service_healthy
+    volumes:
+      - tls-certs:/certs:ro
+    environment:
+      TURNSTONE_TLS_ENABLED: "true"
+      TURNSTONE_TLS_SANS: "server"
+    # Disable healthcheck — server serves HTTPS with mTLS which the
+    # stdlib healthcheck script can't satisfy. The base compose
+    # healthcheck uses plain HTTP which won't work on an HTTPS listener.
+    # TODO: wire healthcheck with client cert from /certs volume
+    healthcheck:
+      disable: true
+
+  # Bridge: mTLS to server + Redis TLS
+  bridge:
+    depends_on:
+      console:
+        condition: service_healthy
+      server:
+        condition: service_started
+      redis:
+        condition: service_healthy
+    volumes:
+      - tls-certs:/certs:ro
+    environment:
+      TURNSTONE_TLS_ENABLED: "true"
+      TURNSTONE_TLS_SANS: "bridge"
+    command:
+      - turnstone-bridge
+      - --server-url=http://server:8080
+      - --redis-host=redis
+      - --redis-port=6379
+      - --heartbeat-ttl=${HEARTBEAT_TTL:-60}
+      - --approval-timeout=${APPROVAL_TIMEOUT:-3600}
+      - --redis-tls
+      - --redis-tls-ca=/certs/ca.pem
+
+  # Channel: Redis TLS
+  channel:
+    depends_on:
+      console:
+        condition: service_healthy
+      redis:
+        condition: service_healthy
+    volumes:
+      - tls-certs:/certs:ro
+    environment:
+      TURNSTONE_TLS_ENABLED: "true"
+      TURNSTONE_TLS_SANS: "channel"
+    command:
+      - sh
+      - -c
+      - >-
+        turnstone-channel
+        --redis-host=redis
+        --redis-port=6379
+        --redis-tls
+        --redis-tls-ca=/certs/ca.pem
+        --http-host=0.0.0.0
+        $${TURNSTONE_DISCORD_GUILD:+--discord-guild $$TURNSTONE_DISCORD_GUILD}
+
+  # Redis: TLS with certs from bootstrap
+  redis:
+    depends_on:
+      tls-init:
+        condition: service_completed_successfully
+    volumes:
+      - tls-certs:/certs:ro
+    command:
+      - sh
+      - -c
+      - |
+        ARGS="--tls-port 6379 --port 0 \
+          --tls-cert-file /certs/certs/redis/cert.pem \
+          --tls-key-file /certs/certs/redis/key.pem \
+          --tls-ca-cert-file /certs/ca.pem \
+          --tls-auth-clients no"
+        if [ -n "$$REDIS_PASSWORD" ]; then
+          ARGS="$$ARGS --requirepass $$REDIS_PASSWORD"
+        fi
+        exec redis-server $$ARGS
+    healthcheck:
+      test: ["CMD-SHELL", "if [ -n \"$$REDIS_PASSWORD\" ]; then redis-cli --tls --cacert /certs/ca.pem -a $$REDIS_PASSWORD ping; else redis-cli --tls --cacert /certs/ca.pem ping; fi"]
+      interval: 5s
+      timeout: 3s
+      retries: 5
+
+volumes:
+  tls-certs:

deploy/openshell/routes.yaml

@@ -0,0 +1,49 @@
+# OpenShell inference routing for Turnstone.
+#
+# When using inference routing, the sandbox process connects to
+# https://inference.local instead of the real LLM API. The OpenShell
+# proxy intercepts, rewrites credentials, and forwards to the backend.
+#
+# This keeps real API keys out of the sandbox entirely — the process
+# only sees opaque placeholder tokens in its environment.
+#
+# Usage:
+#   openshell sandbox run \
+#     --inference-routes deploy/openshell/routes.yaml \
+#     ...
+#
+# Then start turnstone with:
+#   python3 -m turnstone.server --base-url https://inference.local
+#
+# CUSTOMIZE: uncomment one of the provider blocks below.
+
+routes:
+
+  # --- OpenAI ---
+  # - name: inference.local
+  #   endpoint: https://api.openai.com/v1
+  #   model: gpt-5
+  #   provider_type: openai
+  #   protocols:
+  #     - openai_chat_completions
+  #     - model_discovery
+  #   api_key_env: OPENAI_API_KEY
+
+  # --- Anthropic ---
+  # - name: inference.local
+  #   endpoint: https://api.anthropic.com
+  #   model: claude-sonnet-4-6
+  #   provider_type: anthropic
+  #   protocols:
+  #     - anthropic_messages
+  #   api_key_env: ANTHROPIC_API_KEY
+
+  # --- Local model server (vLLM / llama.cpp) ---
+  # No secret resolution needed — local servers typically have no auth.
+  # Omit both api_key and api_key_env to skip credential injection.
+  # - name: inference.local
+  #   endpoint: http://localhost:8000/v1
+  #   model: meta-llama/Llama-3.1-70B-Instruct
+  #   protocols:
+  #     - openai_chat_completions
+  #     - model_discovery

deploy/openshell/turnstone-policy.yaml

@@ -0,0 +1,333 @@
+# OpenShell sandbox policy for Turnstone AI orchestration platform.
+#
+# This policy wraps a turnstone-server process (the primary sandbox target).
+# The bridge, console, and channel gateway are separate processes that would
+# each need their own sandbox with a tailored policy variant.
+#
+# Usage:
+#   openshell sandbox run \
+#     --policy deploy/openshell/turnstone-policy.yaml \
+#     --workdir /project \
+#     -- python3 -m turnstone.server --host 0.0.0.0 --port 8080
+#
+# For inference routing (keeps real API keys out of the sandbox):
+#   openshell sandbox run \
+#     --policy deploy/openshell/turnstone-policy.yaml \
+#     --inference-routes deploy/openshell/routes.yaml \
+#     --workdir /project \
+#     -- python3 -m turnstone.server --host 0.0.0.0 --port 8080 \
+#         --base-url https://inference.local
+#
+# Note: inference.local is intercepted by the OpenShell proxy before
+# network policy evaluation — no network_policies entry is needed for it.
+#
+# Customization points (search for "CUSTOMIZE"):
+#   - OIDC issuer endpoint
+#   - Redis host/port (if not localhost)
+#   - MCP HTTP server endpoints
+#   - Additional tool binaries
+#   - web_fetch domain allowlist
+
+version: 1
+
+# ---------------------------------------------------------------------------
+# Filesystem: Landlock kernel enforcement
+# ---------------------------------------------------------------------------
+# Static — cannot be changed after sandbox creation.
+# include_workdir adds the --workdir path to read_write automatically.
+
+filesystem_policy:
+  include_workdir: true
+
+  read_only:
+    # Python runtime + installed packages (includes turnstone package)
+    - /usr
+    - /lib
+    - /lib64
+    # System essentials
+    - /etc
+    - /proc
+    - /dev/urandom
+    # Turnstone config (read-only — writes go to database)
+    # CUSTOMIZE: adjust if config lives elsewhere
+    - /home/sandbox/.config/turnstone
+
+  read_write:
+    # Working directory is added via include_workdir
+    # Temp files (bash tool scripts, eval workdirs)
+    - /tmp
+    # Shell redirections (2>/dev/null)
+    - /dev/null
+    # SQLite database (default location is workdir, covered by include_workdir)
+    # Logs
+    - /var/log
+
+landlock:
+  # best_effort: degrade gracefully on kernels without Landlock (< 5.13)
+  # Change to hard_requirement for production hardened deployments
+  compatibility: best_effort
+
+# ---------------------------------------------------------------------------
+# Process: privilege separation
+# ---------------------------------------------------------------------------
+
+process:
+  run_as_user: sandbox
+  run_as_group: sandbox
+
+# ---------------------------------------------------------------------------
+# Network: per-endpoint, per-binary allowlisting
+# ---------------------------------------------------------------------------
+# Default-deny. Only listed host:port pairs are reachable.
+# Child processes (MCP servers, bash subcommands) inherit the network
+# namespace — they cannot bypass the proxy.
+
+network_policies:
+
+  # --- LLM API providers ---
+
+  openai_api:
+    name: openai-api
+    endpoints:
+      - host: api.openai.com
+        port: 443
+    binaries:
+      - path: /usr/bin/python3*
+      - path: /usr/local/bin/python3*
+
+  anthropic_api:
+    name: anthropic-api
+    endpoints:
+      - host: api.anthropic.com
+        port: 443
+    binaries:
+      - path: /usr/bin/python3*
+      - path: /usr/local/bin/python3*
+
+  # --- Web search fallback (Tavily) ---
+
+  tavily_api:
+    name: tavily-search
+    endpoints:
+      - host: api.tavily.com
+        port: 443
+    binaries:
+      - path: /usr/bin/python3*
+      - path: /usr/local/bin/python3*
+
+  # --- Skill discovery ---
+
+  skills_registry:
+    name: skills-registry
+    endpoints:
+      - host: skills.sh
+        port: 443
+    binaries:
+      - path: /usr/bin/python3*
+      - path: /usr/local/bin/python3*
+
+  github_api:
+    name: github-api
+    endpoints:
+      - host: api.github.com
+        port: 443
+        protocol: rest
+        tls: terminate
+        enforcement: enforce
+        access: read-only
+      - host: raw.githubusercontent.com
+        port: 443
+    binaries:
+      - path: /usr/bin/python3*
+      - path: /usr/local/bin/python3*
+
+  mcp_registry:
+    name: mcp-registry
+    endpoints:
+      - host: registry.modelcontextprotocol.io
+        port: 443
+        protocol: rest
+        tls: terminate
+        enforcement: enforce
+        access: read-only
+    binaries:
+      - path: /usr/bin/python3*
+      - path: /usr/local/bin/python3*
+
+  # --- OIDC SSO ---
+  # CUSTOMIZE: replace with your identity provider's hostname
+
+  # oidc_provider:
+  #   name: oidc-provider
+  #   endpoints:
+  #     - host: login.example.com
+  #       port: 443
+  #   binaries:
+  #     - path: /usr/bin/python3*
+  #     - path: /usr/local/bin/python3*
+
+  # --- Redis (MQ) ---
+  # CUSTOMIZE: if Redis is not on localhost, add host + allowed_ips.
+  # localhost is blocked by default SSRF protection, so we need allowed_ips.
+
+  redis:
+    name: redis-mq
+    endpoints:
+      - port: 6379
+        allowed_ips:
+          - "127.0.0.1"
+    binaries:
+      - path: /usr/bin/python3*
+      - path: /usr/local/bin/python3*
+
+  # --- Discord (channel integration) ---
+  # Uncomment if using turnstone-channel with Discord adapter.
+
+  # discord:
+  #   name: discord
+  #   endpoints:
+  #     - host: discord.com
+  #       port: 443
+  #     - host: gateway.discord.gg
+  #       port: 443
+  #     - host: cdn.discordapp.com
+  #       port: 443
+  #   binaries:
+  #     - path: /usr/bin/python3*
+  #     - path: /usr/local/bin/python3*
+
+  # --- web_fetch tool: curated domain allowlist ---
+  #
+  # This is the hard tradeoff. Turnstone's web_fetch tool lets the LLM
+  # fetch arbitrary public URLs. OpenShell cannot allow "all HTTPS" —
+  # every domain must be enumerated.
+  #
+  # Strategy: allowlist the domains your workloads actually need.
+  # The web_fetch tool will return a connection error for unlisted domains,
+  # which the LLM handles gracefully (it tells the user it can't reach
+  # that site).
+  #
+  # CUSTOMIZE: add domains your workstreams need to fetch from.
+
+  web_fetch_common:
+    name: web-fetch-common
+    endpoints:
+      # Documentation sites
+      - host: "**.readthedocs.io"
+        port: 443
+      - host: docs.python.org
+        port: 443
+      - host: "**.github.io"
+        port: 443
+      # Package registries (metadata lookups)
+      - host: pypi.org
+        port: 443
+      - host: www.npmjs.com
+        port: 443
+      # Stack Overflow / reference
+      - host: stackoverflow.com
+        port: 443
+      - host: "**.stackexchange.com"
+        port: 443
+      # Wikipedia
+      - host: "**.wikipedia.org"
+        port: 443
+    binaries:
+      - path: /usr/bin/python3*
+      - path: /usr/local/bin/python3*
+
+  # --- MCP HTTP servers ---
+  # CUSTOMIZE: add endpoints for any MCP servers using streamable-http
+  # transport. stdio-transport MCP servers need no network entry (they
+  # communicate via stdin/stdout pipes within the sandbox).
+
+  # mcp_http_servers:
+  #   name: mcp-http
+  #   endpoints:
+  #     - host: mcp.internal.example.com
+  #       port: 443
+  #   binaries:
+  #     - path: /usr/bin/python3*
+  #     - path: /usr/local/bin/python3*
+
+  # --- Bash tool: curl/wget ---
+  # The bash tool can run curl/wget. These inherit the network namespace
+  # so they can only reach allowed endpoints. But they need binary entries
+  # to pass the proxy's identity check.
+
+  bash_network_tools:
+    name: bash-network-tools
+    endpoints:
+      # Mirrors web_fetch_common — curl/wget should have the same reach.
+      - host: "**.readthedocs.io"
+        port: 443
+      - host: docs.python.org
+        port: 443
+      - host: "**.github.io"
+        port: 443
+      - host: pypi.org
+        port: 443
+      - host: www.npmjs.com
+        port: 443
+      - host: stackoverflow.com
+        port: 443
+      - host: "**.stackexchange.com"
+        port: 443
+      - host: "**.wikipedia.org"
+        port: 443
+    binaries:
+      - path: /usr/bin/curl
+      - path: /usr/bin/wget
+
+  # --- Package installation ---
+  # pip install / uv add from the bash tool.
+
+  package_registries:
+    name: package-install
+    endpoints:
+      - host: pypi.org
+        port: 443
+      - host: files.pythonhosted.org
+        port: 443
+      - host: "**.pypi.org"
+        port: 443
+    binaries:
+      - path: /usr/bin/pip*
+      - path: /usr/local/bin/pip*
+      - path: /usr/bin/uv
+      - path: /usr/local/bin/uv
+      - path: /usr/bin/python3*
+      - path: /usr/local/bin/python3*
+
+  # --- Git operations ---
+  # read-only: clone, fetch, pull. No push (L7 enforcement).
+
+  git_operations:
+    name: git-read-only
+    endpoints:
+      - host: github.com
+        port: 443
+        protocol: rest
+        tls: terminate
+        enforcement: enforce
+        rules:
+          - allow:
+              method: GET
+              path: "/**/info/refs*"
+          - allow:
+              method: POST
+              path: "/**/git-upload-pack"
+      - host: gitlab.com
+        port: 443
+        protocol: rest
+        tls: terminate
+        enforcement: enforce
+        rules:
+          - allow:
+              method: GET
+              path: "/**/info/refs*"
+          - allow:
+              method: POST
+              path: "/**/git-upload-pack"
+    binaries:
+      - path: /usr/bin/git

docs/api-reference.md

@@ -554,7 +554,7 @@ Possible `state` values:
 | `error`     | An error occurred                               |
 
 **Fan-out pattern:** Each connected client receives its own bounded queue
-(`maxsize=500`). A dedicated fan-out thread reads from the shared global queue
+(`maxsize=1000`). A dedicated fan-out thread reads from the shared global queue
 and copies each event to every client queue. If a client queue is full, the
 event is silently dropped for that client.
 

docs/architecture.md

@@ -91,7 +91,7 @@ turnstone/
     _config.py        Base ChannelConfig dataclass
     discord/          Discord adapter (bot, cog, views, streaming, config)
   shared_static/      Shared design system (base.css, auth.js, theme.js, toast.js, utils.js, kb.js)
-    katex-0.16.38/    Vendored KaTeX math rendering library (MIT, woff2 fonts)
+    katex-0.16.42/    Vendored KaTeX math rendering library (MIT, woff2 fonts)
   ui/
     colors.py         ANSI color constants with NO_COLOR support
     markdown.py       Streaming terminal markdown renderer (line-buffered)
@@ -100,7 +100,7 @@ turnstone/
       index.html      Single-page app shell (links to CSS and JS)
       style.css       Page-specific UI styles (dashboard, markdown elements, approval blocks)
       renderer.js     Markdown + LaTeX renderer (tables, nested lists, blockquotes, KaTeX math)
-      app.js          Page-specific client-side JavaScript (SSE, workstreams, tool approval)
+      app.js          Split-pane UI (Pane class, binary layout tree, SSE, tool approval)
   tools/
     *.json            15 tool schemas (OpenAI function-calling format + turnstone metadata)
 ```
@@ -353,7 +353,7 @@ remove the tab immediately. Controlled by `--workstream-idle-timeout` (default:
 
 **Workstream eviction at capacity:** When `WorkstreamManager.create()` would
 exceed `max_workstreams` (configurable via `[server].max_workstreams`, default
-10), the oldest IDLE workstream is automatically evicted to make room. The
+50), the oldest IDLE workstream is automatically evicted to make room. The
 `turnstone_workstreams_evicted_total` counter is incremented on each eviction.
 If no IDLE workstream is available the create request fails as before.
 
@@ -375,12 +375,19 @@ non-idle background workstreams above the input prompt.
 ### Web Workstreams
 
 - **Tab bar**: Each workstream renders as a tab with a colored state indicator
-  (CSS `@keyframes pulse` animation per state).
-- **Per-tab SSE**: `connectContentSSE(wsId)` opens
-  `/v1/api/events?ws_id=<id>` for the active tab's event stream.
+  (CSS `@keyframes pulse` animation per state). Clicking a tab switches the
+  focused pane's workstream (or focuses an existing pane showing that ws).
+- **Split panes**: The UI supports tiling multiple workstreams side-by-side or
+  stacked via a binary layout tree. Each `Pane` instance encapsulates its own
+  SSE connection, message area, input, and state (busy, approval, streaming).
+  Split via right-click context menu, pane header buttons, or keyboard
+  (`Ctrl+\`, `Ctrl+Shift+\`). Max 6 panes; no duplicate workstreams across panes.
+  Layout persisted to `localStorage`.
+- **Per-pane SSE**: `Pane.connectSSE(wsId)` opens
+  `/v1/api/events?ws_id=<id>` for each pane's event stream independently.
 - **Global SSE**: `connectGlobalSSE()` opens `/v1/api/events/global` which
   receives `ws_state` broadcasts from all workstreams, used to update tab
-  indicators without switching.
+  indicators and pane headers without switching.
 - **New tab / close**: POST `/v1/api/workstreams/new`, POST `/v1/api/workstreams/close`.
 
 ### Thread Safety
@@ -828,10 +835,16 @@ and are the single source of truth for both backends and Alembic migrations.
 backend = "sqlite"                  # "sqlite" | "postgresql"
 path = ".turnstone.db"              # SQLite file path
 url = ""                            # PostgreSQL connection URL
-pool_size = 5                       # PostgreSQL connection pool size
+pool_size = 2                       # PostgreSQL connection pool size (per process)
 ```
 
-Environment variables: `TURNSTONE_DB_BACKEND`, `TURNSTONE_DB_URL`, `TURNSTONE_DB_PATH`.
+Environment variables: `TURNSTONE_DB_BACKEND`, `TURNSTONE_DB_URL`, `TURNSTONE_DB_PATH`,
+`TURNSTONE_DB_POOL_SIZE`.
+
+The default pool is intentionally small (2 base + 3 overflow = 5 per process)
+because all database operations are short-burst queries that hold connections for
+milliseconds. For clusters with many nodes sharing a PostgreSQL instance, use
+[PgBouncer](pgbouncer.md) in transaction pooling mode.
 
 ### Persistence and Resume
 
@@ -953,6 +966,9 @@ warns if the summary was truncated.
   seeded with `history.replaceState({turnstone: 'dashboard'})` on load. The
   `popstate` listener restores the correct tab or shows the dashboard,
   guarded by `_historyNavigation = true` to prevent re-entrant pushState.
+- **Pane focus**: `mousedown` and `focusin` events on pane containers update
+  `focusedPaneId`. Approval shortcuts (y/n/a) apply to the focused pane.
+  `Ctrl+Alt+Arrow` cycles focus between panes.
 
 ### Eval Resilience
 

docs/console.md

@@ -69,10 +69,11 @@ All reads and writes to the node/workstream map are protected by a single `threa
 
 ### Scale Considerations
 
-- **10,000 workstreams** at ~500 bytes each = ~5 MB in memory
-- **1,000 nodes** polled in parallel with 50 threads at ~100ms each = ~2 second poll cycle
+- **50,000 workstreams** (1,000 nodes × 50 per node) at ~500 bytes each = ~25 MB in memory
+- **1,000 nodes** polled in parallel — fan-out concurrency is configurable via `cluster.node_fan_out_limit` (default 200), yielding 5 batches at ~100ms each = ~0.5 second poll cycle
 - **Filtering and pagination** run in-memory on the full workstream list — sub-millisecond at this scale
-- **SSE fan-out** uses the same per-client queue pattern as the per-node server — backed-up clients get events dropped, not blocking
+- **SSE fan-out** uses per-client queues (2,000 events) — backed-up clients get events dropped, not blocking
+- **Database** — for clusters sharing PostgreSQL, use [PgBouncer](pgbouncer.md) in transaction pooling mode
 
 ---
 
@@ -364,7 +365,7 @@ SSE streams (`/v1/api/events`, `/v1/api/events/global`) are proxied as raw byte
 
 ### Authentication
 
-The proxy forwards the user's JWT to upstream server nodes — it extracts the token from the incoming request's cookie (or `Authorization` header) and adds it as a `Bearer` header on the proxied request.  Since all services share the same `TURNSTONE_JWT_SECRET`, the user's JWT is valid on every node without re-authentication.  The console's own auth middleware also checks proxy routes — `POST` requests to proxy write endpoints (`/v1/api/send`, `/v1/api/approve`, etc.) require `write` scope, preventing read-only tokens from escalating via proxy.  The static `--auth-token` / `proxy_auth_token` is used as a fallback when no user JWT is present.
+The proxy mints a short-lived (5-minute) JWT per request carrying the real user's `user_id`, `scopes`, and `permissions` with `aud: turnstone-server`.  The user's console JWT (`aud: turnstone-console`) cannot be forwarded directly — it would be rejected by the server's audience validation — so the console re-signs a new server-audience JWT from the validated `AuthResult`.  This preserves audit attribution (the upstream server sees the real user, not a service identity) and enforces scope narrowing as defense in depth (a read-only console user's proxied request carries only `read` scope).  The JWT `src` claim is set to `"console-proxy"` for audit traceability.  When no user context is available (auth disabled), the proxy falls back to a `ServiceTokenManager` with service identity `console-proxy`.  The static `--auth-token` / `proxy_auth_token` is used as a final fallback.
 
 ---
 

docs/diagrams/03-core-engine-classes.puml

@@ -250,13 +250,11 @@ class "MCPClientManager" as MCPMgr {
 
 ' ToolSearchManager
 class "ToolSearchManager" as ToolSearchMgr {
-    - _all_tools: list[dict]
     - _always_on: list[dict]
     - _deferred: list[dict]
     - _expanded: dict[str, None]
     - _index: BM25Index
     --
-    + should_activate() → bool
     + get_visible_tools() → list[dict]
     + get_deferred_tools() → list[dict]
     + get_expanded_names() → list[str]

docs/diagrams/06-mq-protocol.puml

@@ -61,6 +61,7 @@ package "Inbound Messages (Client → Bridge)" as InPkg #FFF3E0 {
         + target_node: str = ""
         + initial_message: str = ""
         + skill: str = ""
+        + user_id: str = ""
     }
 
     class CloseWorkstreamMessage {

docs/diagrams/11-console-data-flow.puml

@@ -105,7 +105,7 @@ Server -> CC : get_snapshot()
 CC --> Server : ClusterSnapshot\n(full current state)
 
 Server -> CC : register_listener(queue)
-note right : Per-client queue.Queue(maxsize=500)\nSSE via EventSourceResponse + run_in_executor()
+note right : Per-client queue.Queue(maxsize=2000)\nSSE via EventSourceResponse + run_in_executor()
 
 Server -> Browser : data: {"type":"snapshot",...}\n(full state as first SSE event)
 
@@ -154,7 +154,7 @@ activate Server #FFECB3
 Server -> CC : _pick_best_node() or\nget_node_detail(node_id)
 CC --> Server : node validated
 
-Server -> Server : Build CreateWorkstreamMessage\n{target_node:"nodeA", name:"new-task"}
+Server -> Server : Build CreateWorkstreamMessage\n{target_node:"nodeA", name:"new-task",\nuser_id: from auth_result}
 
 Server -> Redis : RPUSH turnstone:inbound:nodeA\n(directed queue)
 Server --> Browser : {status:"ok", correlation_id:"abc",\ntarget_node:"nodeA"}
@@ -163,7 +163,8 @@ deactivate Server
 note right of Redis
   Bridge on Node-A picks up the
   message from its directed queue,
-  POSTs to /v1/api/workstreams/new,
+  POSTs to /v1/api/workstreams/new
+  (forwarding user_id in payload),
   registers ownership, publishes
   ws_created to cluster channel.
 end note

docs/diagrams/12-deployment.puml

@@ -56,6 +56,26 @@ node "Docker Host" as host {
             end note
         }
 
+        node "postgres (profile: production)" <<pgautoupgrade>> as pg_node {
+            component [PostgreSQL\nport 5432] as postgres
+            note bottom of postgres
+                Healthcheck: pg_isready
+                Volume: postgres-data
+                Required for cluster
+                  and production profiles
+            end note
+        }
+
+        node "pgbouncer (optional)" <<bitnami/pgbouncer>> as pgb_node {
+            component [PgBouncer\nport 6432] as pgbouncer
+            note bottom of pgbouncer
+                pool_mode: transaction
+                Recommended for clusters
+                  > 50 nodes
+                See docs/pgbouncer.md
+            end note
+        }
+
         node "sim (profile: sim)" <<turnstone image>> as sim_node {
             component [turnstone-sim] as sim
             note bottom of sim
@@ -92,6 +112,11 @@ console --> server : HTTP polling + proxy\n(GET /v1/api/dashboard,\nproxy /node/
 
 sim --> redis : Redis protocol\n(queues + pubsub + keys)
 
+' Database connections (production/cluster profiles)
+server ..> pgbouncer : PostgreSQL\n(pool_size=2)
+console ..> pgbouncer : PostgreSQL\n(auth/admin)
+pgbouncer --> postgres : transaction\npooling
+
 ' Environment variables
 note right of host
     **Environment Variables:**
@@ -99,13 +124,17 @@ note right of host
     • OPENAI_API_KEY — API key
     • REDIS_PASSWORD — Redis auth
     • TURNSTONE_AUTH_TOKEN — API auth
+    • TURNSTONE_DB_URL — PostgreSQL URL
+    • POSTGRES_PASSWORD — DB password
 end note
 
 ' Volumes
 database "redis-data" as rv
 database "turnstone-data" as tv
+database "postgres-data" as pv
 
 redis_node --> rv
 server_node --> tv
+pg_node --> pv
 
 @enduml

docs/diagrams/14-storage-architecture.puml

@@ -53,10 +53,10 @@ class "SQLiteBackend" as SQLite <<sqlite>> {
 
 class "PostgreSQLBackend" as PG <<postgres>> {
     -_engine: sa.Engine
-    +__init__(url: str, pool_size: int)
+    +__init__(url: str, pool_size: int = 2,\n    max_overflow: int = 3)
     --
     tsvector + ILIKE search
-    Connection pooling
+    Connection pooling (5 max per process)
 }
 
 ' -- Schema --
@@ -151,7 +151,7 @@ note right of Registry
   backend = "sqlite" | "postgresql"
   url = "postgresql+psycopg://..."
   path = ".turnstone.db"
-  pool_size = 5
+  pool_size = 2 (+ 3 overflow)
 end note
 
 note bottom of SQLite
@@ -162,8 +162,9 @@ end note
 
 note bottom of PG
   Production backend.
-  Multi-node / Docker
-  default.
+  Multi-node / Docker default.
+  Use PgBouncer (transaction mode)
+  for clusters > 50 nodes.
 end note
 
 @enduml

docs/diagrams/15-auth-architecture.puml

@@ -176,4 +176,15 @@ note bottom of SH
   Both share JWT signing secret
 end note
 
+note left of JWT
+  **Console Proxy Token Minting**
+  When proxying requests to server nodes:
+  1. Console AuthMiddleware validates user JWT (aud: turnstone-console)
+  2. Proxy mints new JWT (aud: turnstone-server)
+     with real user_id, scopes, permissions
+  3. src: "console-proxy" for audit traceability
+  4. 5-minute expiry (fresh per request)
+  5. Fallback: ServiceTokenManager if no user context
+end note
+
 @enduml

docs/diagrams/26-skills-discovery-architecture.puml

@@ -33,7 +33,7 @@ package "Core Modules" as core #181825 {
 }
 
 package "Session Runtime" as runtime #181825 {
-  rectangle "load_skill tool\nsession.py" as loadtool
+  rectangle "skill tool\nsession.py" as loadtool
   rectangle "set_skill()\nsession.py" as setskill
   rectangle "_load_skills()\nsession.py" as loadskills
 }
@@ -80,6 +80,7 @@ importui --> install : POST (github source)
 ' Annotations
 note right of parser
   YAML frontmatter -> ParsedSkill
+  allowed-tools (standard) -> allowed_tools (internal)
   Anthropic + Hermes tag formats
   Name validation (lowercase+hyphens)
 end note

docs/diagrams/png/11-console-data-flow.png

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a74b4b8b5dbfb1a51a01100b731477968942b01218bad9451a3d5a9cb3003294
-size 411665
+oid sha256:e3f1ad0fcd55eaca3b8ad9c5abc07432803641c54ede9fc93c79df144cf77d1c
+size 407761

docs/diagrams/png/12-deployment.png

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:84524f4bc900708ac8adf081591d336f862830188eb8505e71a0f071b339d923
-size 252599
+oid sha256:09065fef028d05e6df425fd8abefaf5a2ca04b66802f2e3975f289fa597f63ed
+size 309656

docs/diagrams/png/14-storage-architecture.png

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c94556889abb382cd5b818639fc0a4706beef3d9c7a0b4cbedc763943d657dd0
-size 244998
+oid sha256:b047cdc318c505f0f0895a65e14c5cc7552716053055cca57fa0a77db150e618
+size 255458

docs/docker.md

@@ -109,9 +109,12 @@ All configuration is via environment variables in `.env` (copy from `.env.exampl
 |----------|---------|-------------|
 | `TURNSTONE_DB_BACKEND` | `sqlite` | Storage backend: `sqlite` or `postgresql` |
 | `TURNSTONE_DB_URL` | — | Database URL (e.g. `postgresql://user:pass@db:5432/turnstone`). For SQLite, defaults to `/data/.turnstone.db` |
+| `TURNSTONE_DB_POOL_SIZE` | `2` | PostgreSQL connection pool size per process (default: 2 base + 3 overflow = 5 max) |
 
 The database stores workstream history, user accounts, and API tokens. When using JWT auth, a database backend is required for user storage.
 
+> **Large clusters:** Each turnstone process maintains a small connection pool (5 max). At hundreds of nodes this adds up — use [PgBouncer](pgbouncer.md) in transaction pooling mode between turnstone and PostgreSQL.
+
 > **First-time setup:** After deploying with auth enabled, create an initial admin user by running `turnstone-admin create-user` inside the container:
 >
 > ```bash
@@ -151,6 +154,8 @@ POSTGRES_PASSWORD=secret docker compose --profile cluster up
 
 The default `server` and `bridge` also run alongside the cluster nodes (11 total). All nodes are accessible via the console dashboard at `:8090`.
 
+For production clusters beyond ~50 nodes, add PgBouncer between turnstone services and PostgreSQL. See [PgBouncer Connection Pooling](pgbouncer.md) for Docker Compose and Helm configuration.
+
 ## Volumes
 
 | Volume | Mount | Purpose |

docs/eval.md

@@ -2,7 +2,8 @@
 
 `turnstone-eval` is the evaluation and prompt optimization system for turnstone. It
 runs test cases against the LLM, scores tool call sequences against expected
-actions, and optionally uses the model to self-optimize the developer prompt.
+actions, and optionally uses a multi-agent pipeline to optimize the developer
+prompt and tool descriptions.
 
 Source: `turnstone/eval.py`
 
@@ -10,15 +11,24 @@ Source: `turnstone/eval.py`
 
 ## Overview
 
-The system works in an iterative loop:
+The system uses UCB tree search to explore prompt variants:
 
-1. Run each test case N times against the current developer prompt.
-2. Score each run by comparing the actual tool call sequence to expected actions.
-3. If not all tests pass, use the model to rewrite the prompt based on failures.
-4. Repeat until all tests pass or max iterations are reached.
+1. Maintain an **evolution tree** of prompt variants, starting from the initial prompt.
+2. Each iteration, **UCB1 selects** the most promising node to evaluate.
+3. Run each test case N times against the selected prompt.
+4. Score each run by comparing the actual tool call sequence to expected actions.
+5. If not all tests pass, run a **three-phase optimization pipeline**:
+   - Phase 1: Analyst diagnoses semantic failure patterns
+   - Phase 2: Tool optimizer adjusts tool descriptions (when `--optimize-tools`)
+   - Phase 3: Prompt optimizer proposes a child variant (when not `--optimize-tools`)
+6. Add the child to the tree and repeat until all tests pass or max iterations reached.
 
-When optimization is disabled (`--no-optimize`), only step 1 and 2 execute
-(a single iteration).
+This approach (inspired by [Learning to Self-Evolve](https://arxiv.org/abs/2603.18620))
+prevents irrecoverable collapse from bad edits — UCB naturally backtracks to
+high-scoring ancestors instead of following a linear chain.
+
+When optimization is disabled (`--no-optimize`), only steps 2-4 execute
+(a single iteration evaluating the root node).
 
 ---
 
@@ -65,6 +75,7 @@ Test suites are JSON files with this structure:
 | `match_mode`       | no       | `"ordered_subset"` | How to match actual vs expected actions (see Scoring). |
 | `max_turns`        | no       | `10`               | Maximum conversation turns before stopping. |
 | `n_runs`           | no       | suite default or 3 | Per-case override for number of runs. |
+| `holdout`          | no       | `false`            | If `true`, this case is evaluated but excluded from optimizer feedback. Used to measure progress without overfitting. |
 
 ### Expected Action Specs
 
@@ -135,6 +146,7 @@ deterministic, non-interactive execution suitable for automated testing.
 | Stdout          | Normal                  | Suppressed during execution |
 | Tool logging    | Display only            | Structured `tool_call_log`  |
 | System prompt   | Built-in developer prompt | Overridable via constructor |
+| Cancellation    | N/A                     | `_cancelled` event for timeout cleanup |
 
 ### NullUI
 
@@ -156,20 +168,34 @@ def send_headless(
 Runs a complete multi-turn conversation:
 
 1. Appends the user message.
-2. Calls the model API (non-streaming).
-3. If tool calls are returned, executes them (with stdout suppressed) and
+2. Checks `_cancelled` event — stops if set (timeout cleanup).
+3. Calls the model API (non-streaming).
+4. If tool calls are returned, executes them (with stdout suppressed) and
    logs each call to `self.tool_call_log`.
-4. Repeats up to `max_turns` or until the model responds without tool calls.
-5. Returns the tool call log: list of dicts with keys `tool`, `args`,
+5. Repeats up to `max_turns` or until the model responds without tool calls.
+6. Returns the tool call log: list of dicts with keys `tool`, `args`,
    `result` (truncated to 500 chars), and `turn`.
 
 Parallel tool calls are capped at 10 per turn to prevent degenerate repetition.
 
+### Timeout and Cancellation
+
+Each test runs in a `ThreadPoolExecutor(max_workers=1)` with a per-test
+timeout (`--test-timeout`). Each attempt gets its own `OpenAI` client with
+a matching httpx read timeout. On timeout, three layers of defense prevent
+zombie connections:
+
+1. **httpx timeout**: Per-request read timeout aborts the HTTP call and
+   releases the server slot.
+2. **`_cancelled` event**: Prevents the orphan thread from starting new turns.
+3. **`run_client.close()`**: Closes the connection pool to abort any
+   in-flight request.
+
 ### Retry Logic
 
 `send_headless()` is called inside `_run_single_test()` with retry logic:
 3 attempts with exponential backoff (sleep `2^attempt` seconds) on any
-exception. This prevents transient API errors from poisoning eval scores.
+exception. `TimeoutError` is re-raised immediately (no retry).
 
 ---
 
@@ -180,68 +206,175 @@ Each test case runs in isolation:
 1. A fresh temp directory is created.
 2. Setup files are written to the temp directory.
 3. The working directory is changed to the temp directory.
-4. A new `HeadlessSession` is created with the current developer prompt.
-5. `send_headless()` runs the user prompt through the conversation loop.
-6. The tool log is scored against expected actions.
-7. The temp directory is cleaned up.
+4. A per-attempt `OpenAI` client is created with httpx timeout matching `--test-timeout`.
+5. A new `HeadlessSession` is created with the current developer prompt.
+6. `send_headless()` runs the user prompt through the conversation loop.
+7. The tool log is scored against expected actions.
+8. The temp directory is cleaned up.
 
 The memory database is also isolated per test (an ephemeral SQLite database
 in the temp directory) so tests do not pollute each other or the user's
 real memory store.
 
+### Parallel Execution
+
+With `--parallel N` (N > 1), tests run in a `ProcessPoolExecutor` with N
+workers. Each subprocess creates its own `OpenAI` client. This is suitable
+for remote API endpoints but will overwhelm local inference servers. The
+default (`--parallel 1`) runs tests serially.
+
 ---
 
-## Optimization Loop
+## Model Roles
 
-`run_optimization()` is the main entry point for iterative prompt optimization.
+The eval pipeline uses up to five separate model roles, each independently
+configurable. All roles inherit from the test model by default, with a
+cascade chain:
+
+```
+test model (--base-url, --model)
+  └─ optimizer (--optimizer-*)
+       ├─ observer (--observer-*)
+       ├─ analyst (--analyst-*)
+       ├─ diversifier (--diversifier-*)
+       └─ tool optimizer (--tool-optimizer-*)
+```
+
+| Role | Purpose | When it runs |
+|------|---------|--------------|
+| **Test** | The model being evaluated | Every iteration |
+| **Analyst** | Diagnoses semantic failure patterns with tool use | When pass rate < 100% |
+| **Optimizer** | Rewrites the developer prompt | Every iteration (unless `--optimize-tools`) |
+| **Tool optimizer** | Rewrites tool descriptions | When `--optimize-tools` is set |
+| **Observer** | Tunes the optimizer's strategy | Every 3 iterations |
+| **Diversifier** | Generates prompt paraphrases | Once before the loop (when `--diversify N`) |
+
+Typical setup: local model for test, Opus for analyst, Sonnet for
+optimizer/observer/diversifier.
+
+---
+
+## Optimization Pipeline
 
 ### Flow
 
 ```
 for iteration in 0..max_iterations:
-    1. Run all test cases n_runs times with current prompt
-    2. Score and aggregate results
-    3. Save intermediate results to JSON
-    4. If all tests pass -> stop
-    5. Every 3 iterations (at iteration 2, 5, 8, ...):
-       -> Observer reviews optimizer strategy
-       -> Reset prompt to best-performing iteration
-    6. Propose new prompt via optimizer model call
-    7. If prompt unchanged -> stop
-    8. Continue with new prompt
+    1. UCB select → pick the most promising tree node
+    2. Run all test cases n_runs times with selected node's prompt
+    3. Update node score (rolling mean) and visit count
+    4. Save intermediate results + tree state to JSON
+    5. If all tests pass → stop
+    6. Phase 1: Analyst diagnoses semantic failure patterns
+    7. Phase 2 (--optimize-tools only): Tool optimizer adjusts descriptions
+    8. Phase 3 (default only): Prompt optimizer proposes new prompt
+    9. Every 3 iterations: Observer tunes the optimizer's strategy
+   10. Add child node to tree (if prompt or tools changed)
 ```
 
-### Prompt Proposal (`_propose_prompt_modification`)
+### Phase 1: Analyst (`_run_analyst`)
 
-Uses the model to rewrite the developer prompt based on test results:
+A multi-turn agent with `math` (Python) and `bash` tools for computing
+statistics. It receives per-case results with failure classifications and
+produces a structured diagnosis:
 
-- **Input**: Current prompt, test case definitions, per-case results with
-  actual vs expected tool sequences, and a history of the last 3 iterations.
-- **Optimizer system prompt** (`OPTIMIZER_SYSTEM`): Instructs the model to
-  act as a text rewriter. Key guidance includes:
-  - Address critical failure modes (text-only responses, write_file vs edit_file,
-    unnecessary search before create, missing plan calls).
-  - Preserve phrasing that drives 100% pass rate on passing tests.
-  - Use direct imperative style with concrete tool call examples.
-  - Stay within 130% of original prompt length.
-- **Output**: The rewritten prompt text (stripped of reasoning tags and code fences).
+- **Failure patterns**: Shared root causes across failing cases
+- **Success/failure contrast**: What distinguishes passing from failing cases
+- **Consistency signals**: Systematic (0%), flaky (1-79%), marginal (80-99%)
+- **Recommended fixes**: Priority-ordered patterns/examples to add or adjust
 
-### Observer System (`_observe_and_update_optimizer`)
+The analyst is instructed to frame fixes as patterns and examples, not
+imperative rules — this feeds cleaner signal to the optimizer.
 
-Every 3 iterations, a meta-level "observer" reviews the optimizer's strategy:
+In `--optimize-tools` mode, the analyst receives the current tool descriptions
+(with any overrides applied) and focuses on tool confusion and description
+issues rather than system prompt patterns.
 
-- Analyzes the iteration history: score trends, regressions, prompt length changes,
+### Phase 2: Tool Optimizer (`_propose_tool_overrides`)
+
+Runs when `--optimize-tools` is set. Receives the current tool descriptions,
+confusion failures (where the model picked the wrong tool), and the analyst's
+diagnosis. Returns a JSON override dict that modifies tool descriptions.
+
+Overrides are validated against known tool names — only `description` and
+`parameters` changes are accepted (no tool renaming at eval time).
+
+After each iteration, changed descriptions are logged as old → new diffs
+for easy visual inspection.
+
+### Phase 3: Prompt Optimizer (`_propose_prompt_modification`)
+
+Skipped in `--optimize-tools` mode. Receives the current prompt, test
+results with per-case pass rates and deltas from the parent node, and the
+analyst's diagnosis. Returns a rewritten prompt.
+
+The optimizer is instructed to prefer patterns over rules — concrete tool
+chain examples teach better than imperative directives like "ALWAYS" or
+"NEVER." If the current prompt contains rule-heavy language, the optimizer
+is guided to replace it with examples.
+
+### Two Optimization Surfaces
+
+The system supports alternating between two optimization surfaces:
+
+1. **System prompt optimization** (default): Freeze tool descriptions,
+   optimize the developer prompt. Run until scores plateau.
+2. **Tool description optimization** (`--optimize-tools`): Freeze the system
+   prompt, optimize tool descriptions only. Run until scores plateau.
+
+Each surface lifts the floor for the other — tool description improvements
+may unlock system prompt gains that weren't reachable before, and vice versa.
+
+### Observer (`_observe_and_update_optimizer`)
+
+Every 3 iterations, a meta-level observer reviews the optimizer's strategy:
+
+- Analyzes iteration history: score trends, regressions, prompt length changes,
   and diffs between iterations.
-- Summarizes the optimizer's behavioral patterns (list style, header usage, length).
-- Uses `OBSERVER_SYSTEM` to rewrite the optimizer's own system prompt.
+- Detects whether the optimizer is producing rule-heavy or pattern-based output.
+- Rewrites the optimizer's own system prompt to correct course.
 - Rejects degenerate outputs (over 200% of input length).
-- After updating the optimizer prompt, resets the developer prompt to the
-  best-performing iteration so far.
 
-This two-level optimization (optimizer + observer) helps the system escape
-local minima and adjust its rewriting strategy.
+### Prompt Diversification
 
-### Result Persistence
+When `--diversify N` is set, the diversifier generates N paraphrased variants
+of each test case's user prompt before the optimization loop. Each run cycles
+through variants (round-robin), testing robustness across phrasings.
+
+Variants can be cached back to the test suite JSON with `--save-variants`,
+and auto-loaded on subsequent runs even without `--diversify`.
+
+---
+
+## Evolution Tree
+
+The optimization maintains a tree of prompt variants (`EvolutionNode`), where
+each node stores its prompt text, tool overrides, aggregated score, and visit
+count. The root node (ID 0) contains the initial prompt.
+
+**UCB1 selection**: Each iteration picks the node with the highest Upper
+Confidence Bound score: `R_bar + C * sqrt(ln(N) / v)`, where `R_bar` is the
+node's mean score, `N` is total visits across all nodes, `v` is the node's
+visit count, and `C` is the exploration constant (`--explore-constant`,
+default sqrt(2)). Unvisited nodes are always selected first.
+
+### Holdout Cases
+
+Test cases with `"holdout": true` are evaluated every iteration but excluded
+from the optimizer's feedback. This prevents the optimizer from overfitting
+to specific test cases. Node scores are computed from holdout cases only
+(when present). If fewer than 2 non-holdout cases remain, holdout is disabled.
+
+### Improvement-Based Feedback
+
+The optimizer sees delta scores (`delta=+20%`) alongside absolute pass rates,
+showing how each case improved relative to the parent node's evaluation. This
+provides a cleaner signal than absolute scores alone — the optimizer can
+distinguish beneficial edits from harmful ones regardless of starting point.
+
+---
+
+## Result Persistence
 
 After each iteration, results are written to the output JSON file. The
 structure is:
@@ -251,9 +384,15 @@ structure is:
   "meta": {
     "model": "model-name",
     "base_url": "http://localhost:8000/v1",
+    "optimizer_model": "claude-opus-4-6",
+    "observer_model": "claude-opus-4-6",
     "started": "2025-01-01T00:00:00",
     "test_suite": "tests.json",
-    "n_runs_default": 3
+    "n_runs_default": 3,
+    "explore_constant": 1.414,
+    "holdout_ids": [],
+    "diversify": 10,
+    "prompt_variants": {"case_id": ["variant1", "variant2"]}
   },
   "iterations": [
     {
@@ -261,7 +400,11 @@ structure is:
       "prompt": "the developer prompt used",
       "prompt_diff": null,
       "optimizer_system": "the optimizer system prompt",
+      "analyst": "analyst diagnosis output",
+      "tool_overrides": {"bash": {"description": "..."}},
       "timestamp": "2025-01-01T00:01:00",
+      "tree_node_id": 0,
+      "tree_child_id": 1,
       "cases": {
         "test_name": {
           "runs": [
@@ -287,9 +430,21 @@ structure is:
         "overall_pass_rate": 0.8,
         "overall_avg_score": 0.87,
         "json_dumps": 0,
-        "per_case_pass_rates": {"test_name": 1.0, ...}
+        "per_case_pass_rates": {"test_name": 1.0}
       }
     }
+  ],
+  "tree": [
+    {
+      "node_id": 0,
+      "parent_id": null,
+      "prompt": "initial prompt",
+      "tool_overrides": {},
+      "score": 0.85,
+      "visit_count": 3,
+      "children": [1, 2],
+      "iteration": 0
+    }
   ]
 }
 ```
@@ -306,26 +461,57 @@ turnstone-eval tests.json                          # evaluate + optimize
 turnstone-eval tests.json --no-optimize            # evaluate only (single iteration)
 turnstone-eval tests.json --n-runs 5 --max-iter 10 # more thorough evaluation
 turnstone-eval tests.json --prompt custom.txt      # start from a custom prompt
+turnstone-eval tests.json --optimize-tools         # optimize tool descriptions only
+turnstone-eval tests.json --diversify 10           # test with prompt variants
 turnstone-eval tests.json -v                       # verbose per-turn logging
 ```
 
+### Multi-model setup (local test model, cloud optimizer)
+
+```
+turnstone-eval tests.json \
+  --base-url http://localhost:8000/v1 \
+  --optimizer-base-url https://api.anthropic.com \
+  --optimizer-model claude-sonnet-4-6 \
+  --analyst-model claude-opus-4-6
+```
+
 ### All Options
 
-| Flag                | Default                       | Description |
-|---------------------|-------------------------------|-------------|
-| `test_file`         | (positional, required)        | Path to test cases JSON file. |
-| `--base-url`        | `http://localhost:8000/v1`    | API base URL. |
-| `--model`           | auto-detect                   | Model name. Auto-detected from the API if not specified. |
-| `--prompt`          | turnstone built-in prompt         | Path to initial prompt text file. |
-| `--n-runs`          | from tests.json or 3          | Number of runs per test case. |
-| `--max-iter`        | 5                             | Maximum optimization iterations. |
-| `--no-optimize`     | false                         | Run evaluation only (sets max-iter to 1). |
-| `--temperature`     | 0.7                           | Sampling temperature. |
-| `--max-tokens`      | 32768                         | Max completion tokens. |
-| `--reasoning-effort` | `medium`                     | Reasoning effort: `low`, `medium`, or `high`. |
-| `--context-window`  | 131072                        | Context window size. |
-| `--output`          | `eval_results.json`           | Output results file path. |
-| `-v`, `--verbose`   | false                         | Show detailed per-turn logging (API calls, tool args, results). |
+| Flag                    | Default                    | Description |
+|-------------------------|----------------------------|-------------|
+| `test_file`             | (positional, required)     | Path to test cases JSON file. |
+| `--base-url`            | `http://localhost:8000/v1` | API base URL for the test model. |
+| `--model`               | auto-detect                | Model name. Auto-detected from the API if not specified. |
+| `--prompt`              | turnstone built-in prompt  | Path to initial prompt text file. |
+| `--n-runs`              | from tests.json or 3       | Number of runs per test case. |
+| `--max-iter`            | 5                          | Maximum optimization iterations. |
+| `--no-optimize`         | false                      | Run evaluation only (sets max-iter to 1). |
+| `--temperature`         | 0.7                        | Sampling temperature. |
+| `--max-tokens`          | 32768                      | Max completion tokens. |
+| `--reasoning-effort`    | `medium`                   | Reasoning effort: `low`, `medium`, or `high`. |
+| `--context-window`      | 131072                     | Context window size. |
+| `--output`              | `eval_results.json`        | Output results file path. |
+| `-v`, `--verbose`       | false                      | Show detailed per-turn logging. |
+| `--explore-constant`    | 1.414 (sqrt(2))            | UCB exploration constant C. |
+| `--test-timeout`        | 300                        | Per-test timeout in seconds. |
+| `--suite-timeout`       | 0 (unlimited)              | Total suite timeout in seconds. |
+| `--no-fast-fail`        | false                      | Disable early termination on all-zero initial runs. |
+| `--parallel`            | 1 (serial)                 | Parallel workers (0=auto, N=use N workers). |
+| `--optimizer-model`     | same as `--model`          | Model for prompt optimization. |
+| `--optimizer-base-url`  | same as `--base-url`       | Base URL for optimizer model. |
+| `--observer-model`      | same as optimizer           | Model for meta-optimization (observer). |
+| `--observer-base-url`   | same as optimizer           | Base URL for observer model. |
+| `--analyst-model`       | same as optimizer           | Model for failure analysis. |
+| `--analyst-base-url`    | same as optimizer           | Base URL for analyst model. |
+| `--diversify`           | 0 (disabled)               | Generate N prompt variants per test case. |
+| `--diversifier-model`   | same as optimizer           | Model for prompt diversification. |
+| `--diversifier-base-url`| same as optimizer           | Base URL for diversifier model. |
+| `--save-variants`       | false                      | Save generated variants back to test suite JSON. |
+| `--optimize-tools`      | false                      | Optimize tool descriptions only (freeze system prompt). |
+| `--tool-optimizer-model` | same as optimizer          | Model for tool description optimization. |
+| `--tool-optimizer-base-url` | same as optimizer      | Base URL for tool optimizer model. |
+| `--save-tools`          | false                      | Write optimized tool descriptions back to `turnstone/tools/*.json`. |
 
 ### Precedence for n_runs
 

docs/governance.md

@@ -70,7 +70,7 @@ etc.) since workstream templates were merged into the skills system in v0.8.0.
   `{{node_id}}` (server node ID). Unrecognized placeholders are kept as-is.
 - **Runtime switching**: `/template <name>` to switch, `/template clear` to revert
   to defaults, `/template` to show current. Persisted across resume.
-- **Model-driven loading**: The `load_skill` built-in tool lets the model
+- **Model-driven loading**: The `skill` built-in tool lets the model
   discover and activate skills mid-conversation. `search` action finds skills
   by query (auto-approved); `load` action activates by name (requires user
   approval since it changes session behavior). Main session only.
@@ -83,11 +83,15 @@ etc.) since workstream templates were merged into the skills system in v0.8.0.
   precedence on name collision. MCP-synced content updates reset `is_default` to
   prevent compromised servers from injecting defaults. Admin UI shows origin badge
   and disables edit/delete for MCP-sourced skills.
+- **Spec fields**: Skills support the full Agent Skills standard frontmatter:
+  `name`, `description`, `license`, `compatibility`, `metadata` (author, version),
+  `allowed-tools`. The `license` and `compatibility` fields are preserved on import
+  and editable in the admin UI. See https://agentskills.io/specification.
 - **Security scanning**: Skills are automatically scanned at creation and update
   time. The scanner evaluates four risk axes: content risk (command execution,
   data exfiltration), supply chain risk (pipe-to-shell, transitive installs),
   vulnerability risk (prompt injection, insecure credentials), and declared
-  capability risk (from `allowed_tools`). Results populate the `scan_status`
+  capability risk (from `allowed-tools` in SKILL.md). Results populate the `scan_status`
   (safe/low/medium/high/critical) and `scan_report` (JSON breakdown) columns.
   These fields are system-managed and cannot be overwritten via the admin API.
 - **Discovery**: External skills can be discovered and installed from registries:
@@ -100,6 +104,20 @@ etc.) since workstream templates were merged into the skills system in v0.8.0.
     Discovery view has search bar, result cards, and "Import from GitHub" modal.
   - SDK: `discover_skills(q)` and `install_skill(source, skill_id=..., url=...)`
     on both Python and TypeScript console clients.
+- **Runtime config on installed skills**: Installed (readonly) skills can have
+  their runtime configuration edited — model, temperature, reasoning effort,
+  token budget, max tokens, agent max turns, auto-approve, allowed tools,
+  and enabled flag. The server restricts updates to these fields only via
+  `_SKILL_RUNTIME_CONFIG_FIELDS` filtering; spec/content fields (name,
+  description, tags, license, compatibility, content, activation) remain
+  immutable. The admin UI shows "Save Config" instead of "Save" for these
+  skills. Audit action: `skill.update.config`.
+- **Admin UI**: Create/Edit skill modals use a two-column spec manifest layout
+  (left: Identity / Manifest / Deployment; right: Skill Content editor with
+  monospace font). Runtime Config is a collapsible 3-column grid below.
+  License uses an SPDX identifier dropdown (MIT, Apache-2.0, GPL-3.0, etc.).
+  Installed skills show a cyan origin badge with source URL, spec fields are
+  disabled, and all collapsible sections auto-expand in view mode.
 
 ### Usage Tracking
 

docs/judge.md

@@ -299,7 +299,7 @@ four independent risk axes:
    obfuscation, download-execute chains, executable URLs from untrusted domains
 3. **Vulnerability risk** — prompt injection patterns, insecure credential
    handling, third-party content exposure (indirect prompt injection surface)
-4. **Declared capability risk** — parsed from the skill's `allowed_tools` field.
+4. **Declared capability risk** — parsed from `allowed-tools` in the skill's SKILL.md.
    `Bash(*)` (unrestricted shell) is high risk. `Bash(git:*)` is low.
    Read-only tools are safe.
 

docs/openshell.md

@@ -0,0 +1,288 @@
+# OpenShell Sandbox Integration
+
+Turnstone can run inside an [OpenShell](https://github.com/NVIDIA/OpenShell)
+sandbox for kernel-enforced security boundaries around tool execution. OpenShell
+provides four layers of defense that Turnstone's application-level safety model
+does not cover:
+
+| Layer | Mechanism | What it prevents |
+|-------|-----------|------------------|
+| Filesystem | Landlock | Writes to `/etc`, `~/.ssh`, system paths |
+| Network | Network namespace + seccomp + HTTP CONNECT proxy | Connections to unlisted hosts |
+| Process | `setuid` drop + verification | Privilege escalation to root |
+| Credentials | Proxy-level secret resolution | API keys in sandbox memory |
+
+Turnstone's own safety layers (human approval, intent judge, tool policies,
+output guard) remain active inside the sandbox and handle threats at the semantic
+level -- what the LLM *means* to do with its legitimate access.
+
+> See also: [Security and Authentication](security.md),
+> [Intent Validation](judge.md), [Governance](governance.md)
+
+---
+
+## Quick Start
+
+```bash
+# Run turnstone-server in an OpenShell sandbox
+openshell sandbox run \
+  --policy deploy/openshell/turnstone-policy.yaml \
+  --workdir /path/to/project \
+  -- python3 -m turnstone.server --host 0.0.0.0 --port 8080
+```
+
+With inference routing (API keys never enter the sandbox):
+
+```bash
+openshell sandbox run \
+  --policy deploy/openshell/turnstone-policy.yaml \
+  --inference-routes deploy/openshell/routes.yaml \
+  --workdir /path/to/project \
+  -- python3 -m turnstone.server --host 0.0.0.0 --port 8080 \
+      --base-url https://inference.local
+```
+
+The `inference.local` hostname is intercepted by the OpenShell proxy before
+network policy evaluation -- no network policy entry is needed for it.
+
+---
+
+## Policy Files
+
+### `deploy/openshell/turnstone-policy.yaml`
+
+The main sandbox policy. Covers filesystem, process, and network rules.
+
+### `deploy/openshell/routes.yaml`
+
+Inference routing configuration. Maps `inference.local` to real LLM API
+backends. Uncomment and configure the provider(s) you use.
+
+---
+
+## Filesystem Policy
+
+The policy uses Landlock (Linux 5.13+) for kernel-enforced filesystem access
+control. Paths are locked at sandbox creation and cannot be changed at runtime.
+
+| Path | Access | Purpose |
+|------|--------|---------|
+| `--workdir` | read-write | Project files (auto-added via `include_workdir`) |
+| `/tmp` | read-write | Bash tool temp scripts, eval workdirs |
+| `/dev/null` | read-write | Shell redirections (`2>/dev/null`) |
+| `/var/log` | read-write | Log files |
+| `/usr`, `/lib`, `/lib64` | read-only | Python runtime, installed packages |
+| `/etc` | read-only | System config, SSL certificates |
+| `/proc`, `/dev/urandom` | read-only | Process info, entropy |
+| `~/.config/turnstone` | read-only | Config file (writes go to database) |
+
+Landlock runs in `best_effort` mode by default -- degrades gracefully on kernels
+without Landlock support. Set `compatibility: hard_requirement` for production
+hardened deployments.
+
+---
+
+## Network Policy
+
+Default-deny. Only explicitly listed host:port pairs are reachable. All child
+processes (MCP servers, bash commands, grep) inherit the network namespace and
+cannot bypass the proxy.
+
+### Included endpoints
+
+| Policy | Hosts | Purpose |
+|--------|-------|---------|
+| `openai_api` | `api.openai.com` | OpenAI LLM API |
+| `anthropic_api` | `api.anthropic.com` | Anthropic LLM API |
+| `tavily_api` | `api.tavily.com` | Web search fallback |
+| `skills_registry` | `skills.sh` | Skill discovery |
+| `github_api` | `api.github.com` (read-only L7), `raw.githubusercontent.com` | Skill fetch, GitHub API |
+| `mcp_registry` | `registry.modelcontextprotocol.io` (read-only L7) | MCP server discovery |
+| `redis` | `127.0.0.1:6379` | Message queue |
+| `web_fetch_common` | readthedocs, python docs, GitHub Pages, PyPI, npm, Stack Overflow, Wikipedia | Curated web_fetch domains |
+| `bash_network_tools` | Same as `web_fetch_common` | curl/wget from bash tool |
+| `package_registries` | `pypi.org`, `files.pythonhosted.org` | pip/uv package installs |
+| `git_operations` | `github.com`, `gitlab.com` (L7: clone/fetch only, no push) | Git read-only operations |
+
+### L7 enforcement
+
+Endpoints marked with `protocol: rest` and `tls: terminate` get HTTP-level
+inspection. The proxy TLS-terminates using an ephemeral per-sandbox CA, parses
+each request, and evaluates method + path against the rules.
+
+The `github_api`, `mcp_registry`, and `git_operations` policies use L7
+enforcement:
+
+- **GitHub API / MCP Registry**: `access: read-only` -- only GET, HEAD, OPTIONS
+  allowed
+- **Git operations**: explicit rules allowing only `info/refs` (GET) and
+  `git-upload-pack` (POST) -- clone and fetch work, push is blocked
+
+### Commented-out sections
+
+The policy includes commented blocks for optional integrations. Uncomment and
+configure as needed:
+
+- **OIDC** -- add your identity provider's hostname
+- **Discord** -- `discord.com`, `gateway.discord.gg`, `cdn.discordapp.com`
+- **MCP HTTP servers** -- any MCP servers using streamable-http transport
+
+---
+
+## Customizing the Domain Allowlist
+
+The `web_fetch` tool lets the LLM fetch arbitrary public URLs, but OpenShell
+cannot allow "all HTTPS" -- bare wildcard hosts are rejected by policy
+validation. Instead, the policy ships with a curated set of common reference
+domains.
+
+To add domains your workloads need:
+
+```yaml
+# In turnstone-policy.yaml, under web_fetch_common.endpoints:
+      - host: docs.example.com
+        port: 443
+
+# Also add to bash_network_tools.endpoints if curl/wget should reach it:
+      - host: docs.example.com
+        port: 443
+```
+
+Wildcard patterns are supported:
+
+- `*.example.com` -- matches one subdomain level (e.g. `api.example.com`)
+- `**.example.com` -- matches any depth (e.g. `deep.sub.example.com`)
+
+Unlisted domains return connection errors, which the LLM handles gracefully by
+telling the user it cannot reach that site.
+
+---
+
+## Inference Routing
+
+Inference routing keeps real API keys completely outside the sandbox. The
+sandbox process only sees opaque placeholder tokens in its environment
+(`openshell:resolve:env:ANTHROPIC_API_KEY`). The proxy rewrites these to real
+credentials on the wire before forwarding to the upstream API.
+
+### Setup
+
+1. Edit `deploy/openshell/routes.yaml` -- uncomment your provider:
+
+```yaml
+routes:
+  # OpenAI
+  - name: inference.local
+    endpoint: https://api.openai.com/v1
+    model: gpt-5
+    provider_type: openai
+    protocols:
+      - openai_chat_completions
+      - model_discovery
+    api_key_env: OPENAI_API_KEY
+
+  # Or Anthropic
+  - name: inference.local
+    endpoint: https://api.anthropic.com
+    model: claude-sonnet-4-6
+    provider_type: anthropic
+    protocols:
+      - anthropic_messages
+    api_key_env: ANTHROPIC_API_KEY
+```
+
+2. Start with `--inference-routes` and point turnstone at `inference.local`:
+
+```bash
+openshell sandbox run \
+  --inference-routes deploy/openshell/routes.yaml \
+  --base-url https://inference.local \
+  ...
+```
+
+3. When inference routing is active, the `openai_api` and `anthropic_api`
+   network policies can be removed from the sandbox policy -- the proxy handles
+   LLM traffic on a separate code path that bypasses OPA entirely.
+
+### Local model servers
+
+For local servers (vLLM, llama.cpp) with no authentication, omit both
+`api_key` and `api_key_env` from the route config. No credential resolution
+is needed.
+
+---
+
+## MCP Server Subprocesses
+
+MCP servers using stdio transport are spawned as child processes of turnstone.
+They automatically inherit all sandbox constraints:
+
+- **Network namespace** -- kernel-level, cannot be bypassed
+- **Landlock filesystem** -- kernel-level, cannot be relaxed
+- **Seccomp socket filter** -- kernel-level, inherited on fork
+
+No per-subprocess policy entries are needed for these constraints. However, if
+an MCP server makes outbound network requests (through the proxy), its binary
+must appear in a `binaries[]` entry for the relevant network policy. The proxy
+identifies the requesting process via `/proc/<pid>/exe` (not `argv[0]`, which
+is spoofable).
+
+Example for a Python-based MCP server that calls an external API:
+
+```yaml
+  mcp_external_api:
+    name: mcp-external
+    endpoints:
+      - host: api.example.com
+        port: 443
+    binaries:
+      - path: /usr/bin/python3*
+      - path: /usr/local/bin/python3*
+```
+
+MCP servers using streamable-http transport are remote -- they need a network
+policy entry for their host:port but no binary entry (the Python process making
+the HTTP call is already covered by the standard `python3*` binary entries).
+
+---
+
+## Security Model: Which Layer Enforces What
+
+```
+                      OpenShell (infrastructure)          Turnstone (application)
+                     ─────────────────────────────     ──────────────────────────────
+Filesystem access     Landlock kernel enforcement       (no enforcement)
+Network egress        Netns + seccomp + proxy + OPA     SSRF check on web_fetch
+Credentials           Placeholder injection + proxy     Output guard redaction
+Privilege level       setuid drop + verification        (no enforcement)
+Tool semantics        (no visibility)                   Heuristic + LLM judge
+Tool policies         (no visibility)                   fnmatch admin policies
+Prompt injection      (no visibility)                   Output guard detection
+Human approval        (no visibility)                   Approval gate + "always"
+```
+
+OpenShell constrains what the process can physically reach. Turnstone constrains
+what the LLM does with its legitimate access. Neither layer is sufficient alone:
+
+- Without OpenShell: a bash command can `curl` secrets to any endpoint, write to
+  `/etc/crontab`, or read `~/.ssh/id_rsa` -- all gated only by human approval
+- Without Turnstone: the LLM can `rm -rf` the entire workdir, run destructive
+  commands, or consume prompt injection payloads -- all within the sandbox's
+  allowed scope
+
+---
+
+## Hardening Checklist
+
+For production deployments:
+
+- [ ] Set `landlock.compatibility: hard_requirement`
+- [ ] Enable inference routing (removes API keys from sandbox)
+- [ ] Remove `openai_api`/`anthropic_api` network policies when using inference
+      routing (traffic goes through the router, not direct)
+- [ ] Review and trim `web_fetch_common` domains to your actual needs
+- [ ] Remove `package_registries` policy if pip/uv installs are not needed
+- [ ] Add your OIDC provider endpoint if using SSO
+- [ ] Set Redis `allowed_ips` to your actual Redis host if not localhost
+- [ ] Consider removing `bash_network_tools` entirely if bash should not have
+      network access

docs/pgbouncer.md

@@ -0,0 +1,200 @@
+# PgBouncer Connection Pooling
+
+Turnstone cluster deployments share a single PostgreSQL instance across
+all server nodes, bridge processes, and the console. Each process
+maintains a small connection pool (2 base + 3 overflow = 5 max). At
+scale this adds up — a 100-node cluster opens up to 500 connections,
+and a 1000-node cluster up to 5,000.
+
+PostgreSQL's default `max_connections` is 100, and each real connection
+allocates ~5–10 MB of backend memory. PgBouncer sits between turnstone
+and PostgreSQL, multiplexing thousands of lightweight client connections
+down to a small number of real database connections.
+
+---
+
+## Why PgBouncer works well with turnstone
+
+All turnstone database operations are short-burst queries: acquire a
+connection, execute 1–3 statements, commit, release. No operation holds
+a connection for more than a few milliseconds. This makes **transaction
+pooling mode** ideal — PgBouncer assigns a real connection only for the
+duration of each transaction, then returns it to the pool.
+
+| Cluster size | Client connections (max) | PgBouncer server connections needed |
+|--------------|------------------------|-------------------------------------|
+| 10 nodes     | 50                     | 10–20                               |
+| 100 nodes    | 500                    | 20–40                               |
+| 500 nodes    | 2,500                  | 30–60                               |
+| 1,000 nodes  | 5,000                  | 40–80                               |
+
+The server connection count stays low because most client connections
+are idle at any given moment.
+
+---
+
+## Docker Compose
+
+Add PgBouncer between turnstone services and PostgreSQL:
+
+```yaml
+services:
+  pgbouncer:
+    image: bitnami/pgbouncer:latest
+    environment:
+      POSTGRESQL_HOST: postgres
+      POSTGRESQL_PORT: "5432"
+      POSTGRESQL_DATABASE: turnstone
+      POSTGRESQL_USERNAME: ${POSTGRES_USER:-turnstone}
+      POSTGRESQL_PASSWORD: ${POSTGRES_PASSWORD:?}
+      PGBOUNCER_POOL_MODE: transaction
+      PGBOUNCER_DEFAULT_POOL_SIZE: "40"
+      PGBOUNCER_MAX_CLIENT_CONN: "5000"
+      PGBOUNCER_MAX_DB_CONNECTIONS: "80"
+      PGBOUNCER_SERVER_IDLE_TIMEOUT: "300"
+    ports:
+      - "6432:6432"
+    networks:
+      - turnstone-net
+    depends_on:
+      postgres:
+        condition: service_healthy
+    healthcheck:
+      test: ["CMD", "pg_isready", "-h", "127.0.0.1", "-p", "6432"]
+      interval: 5s
+      timeout: 3s
+      retries: 5
+```
+
+Then point turnstone services at PgBouncer instead of PostgreSQL
+directly by changing the `DATABASE_URL` (or `TURNSTONE_DB_URL`):
+
+```bash
+# Before (direct)
+TURNSTONE_DB_URL=postgresql://turnstone:secret@postgres:5432/turnstone
+
+# After (via PgBouncer)
+TURNSTONE_DB_URL=postgresql://turnstone:secret@pgbouncer:6432/turnstone
+```
+
+---
+
+## Helm / Kubernetes
+
+Add a PgBouncer deployment or use a Helm chart like
+[bitnami/pgbouncer](https://github.com/bitnami/charts/tree/main/bitnami/pgbouncer).
+
+In `values.yaml`, point the database at PgBouncer:
+
+```yaml
+database:
+  backend: postgresql
+  external:
+    host: pgbouncer
+    port: 6432
+    database: turnstone
+    username: turnstone
+    existingSecret: turnstone-db-secret
+```
+
+PgBouncer configuration:
+
+```yaml
+pgbouncer:
+  poolMode: transaction
+  defaultPoolSize: 40
+  maxClientConn: 5000
+  maxDbConnections: 80
+```
+
+---
+
+## Configuration reference
+
+| PgBouncer setting | Recommended | Notes |
+|-------------------|-------------|-------|
+| `pool_mode` | `transaction` | Required — turnstone uses short-burst queries with no session state |
+| `default_pool_size` | 40 | Real PostgreSQL connections per database. Start here, increase if you see `no more connections allowed` |
+| `max_client_conn` | 5000 | Upper bound on client connections. Set to `cluster_nodes × 5` |
+| `max_db_connections` | 80 | Hard cap on real connections to PostgreSQL. Keep below PG `max_connections` minus headroom for admin/monitoring |
+| `server_idle_timeout` | 300 | Close idle server connections after 5 minutes |
+| `server_lifetime` | 3600 | Recycle server connections after 1 hour |
+
+On the PostgreSQL side:
+
+| PostgreSQL setting | Recommended | Notes |
+|--------------------|-------------|-------|
+| `max_connections` | 100 | Default is fine — PgBouncer is the only client. Set higher than `max_db_connections` to leave room for admin connections |
+| `shared_buffers` | 25% of RAM | Standard PostgreSQL tuning |
+
+---
+
+## Turnstone pool settings
+
+Each turnstone process maintains its own SQLAlchemy connection pool to
+PgBouncer (which then multiplexes to PostgreSQL):
+
+| Environment variable | Default | Description |
+|---------------------|---------|-------------|
+| `TURNSTONE_DB_POOL_SIZE` | 2 | Base pool size per process |
+| `TURNSTONE_DB_BACKEND` | sqlite | Set to `postgresql` for cluster deployments |
+| `TURNSTONE_DB_URL` | — | Connection URL (point at PgBouncer, not PostgreSQL directly) |
+
+The default pool of 2 + 3 overflow = 5 connections per process is
+intentionally small to support large clusters. You should not need to
+increase this — turnstone's database operations are all short-burst
+context-managed queries that hold connections for milliseconds.
+
+SQLAlchemy `pool_pre_ping` is enabled, so stale connections (e.g. after
+PgBouncer restarts) are automatically detected and replaced.
+
+---
+
+## Monitoring
+
+PgBouncer exposes stats via its admin console (connect to
+PgBouncer port with user `pgbouncer`):
+
+```sql
+-- Active and waiting clients
+SHOW POOLS;
+
+-- Per-database stats
+SHOW STATS;
+
+-- Current client connections
+SHOW CLIENTS;
+```
+
+Key metrics to watch:
+
+- **`cl_active`** — clients with a server connection assigned. Should be
+  well below `max_db_connections`.
+- **`cl_waiting`** — clients waiting for a server connection. Sustained
+  non-zero values mean you need more `default_pool_size`.
+- **`sv_active`** — active server (PostgreSQL) connections. Should stay
+  below PostgreSQL `max_connections`.
+
+---
+
+## Troubleshooting
+
+**"no more connections allowed (max_client_conn)"** — PgBouncer is
+rejecting new client connections. Increase `max_client_conn` to match
+your cluster size × 5.
+
+**"no more connections allowed (max_db_connections)"** — PgBouncer
+cannot open more connections to PostgreSQL. Increase
+`max_db_connections` and ensure PostgreSQL `max_connections` is higher.
+
+**Connections timing out on startup** — If all nodes start
+simultaneously, the burst of initial connections (migrations, health
+checks) can temporarily exceed the pool. PgBouncer queues excess
+clients by default — this resolves itself within seconds.
+
+**Prepared statements not supported** — PgBouncer in `transaction` mode
+does not support prepared statements. Turnstone's SQLAlchemy layer does
+not use server-side prepared statements by default, so this is not an
+issue.
+
+See also: [Docker deployment](docker.md) · [Security](security.md)

docs/security.md

@@ -457,14 +457,30 @@ without any database.
 ### Proxy auth forwarding
 
 When the console proxies requests to server nodes (via `/node/{id}/...`
-routes), it uses a dedicated **service proxy token** with
-`aud: turnstone-server` and `write` scope.  The user's console JWT
-(which has `aud: turnstone-console`) is **not** forwarded — it would be
-rejected by the server's audience validation.
+routes), it mints a **short-lived user-scoped JWT** with
+`aud: turnstone-server` carrying the real user's `user_id`, `scopes`,
+and `permissions`.  The user's console JWT (which has
+`aud: turnstone-console`) is **not** forwarded directly — it would be
+rejected by the server's audience validation.  Instead, the console
+re-signs a new JWT targeted at the server audience.
 
-The proxy token is managed by a `ServiceTokenManager` that auto-rotates
-1-hour JWTs, refreshing at 80% of lifetime.  If `--auth-token` is
-provided, that static token is used instead.
+Each proxied request gets a fresh JWT (5-minute expiry).  This ensures:
+
+- **Audit attribution** — the upstream server records the real user in
+  `ctx_user_id` and audit events, not a generic service identity.
+- **Scope narrowing** — a read-only console user's proxied request
+  carries only `read` scope, not the full `{read, write, approve}` set.
+  The server enforces this as defense in depth.
+- **Permission forwarding** — granular RBAC permissions from the
+  console JWT are carried through to the server.
+
+The JWT `src` claim is set to `"console-proxy"`, allowing servers to
+distinguish proxied requests from direct logins in audit logs.
+
+When no user context is available (auth disabled, or internal requests),
+the proxy falls back to a `ServiceTokenManager` with service identity
+`console-proxy` and full scopes.  If `--auth-token` is provided, that
+static token is used as a final fallback.
 
 ### Service-to-service authentication
 
@@ -475,7 +491,7 @@ auto-rotating JWTs when communicating with server nodes:
 |---------|----------|-------|----------|---------|
 | 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 |
+| Console proxy (fallback) | `console-proxy` | `approve` | `turnstone-server` | Proxied API calls when no user context |
 | Channel notify | `system` | `write` | `turnstone-channel` | Notification delivery to channel gateway |
 
 Service tokens use 1-hour expiry with automatic refresh via
@@ -483,6 +499,17 @@ Service tokens use 1-hour expiry with automatic refresh via
 httpx event hooks to ensure rotated tokens are picked up on SSE
 reconnects.
 
+### User identity in MQ-dispatched workstreams
+
+When the console creates a workstream via MQ (the normal path), the
+authenticated user's `user_id` is embedded in the
+`CreateWorkstreamMessage`.  The bridge forwards this `user_id` in the
+HTTP payload when calling the server's `POST /v1/api/workstreams/new`.
+The server accepts a `user_id` from the request body **only when the
+caller is a trusted service** — identified by `token_source` matching
+`bridge`, `console-proxy`, or `console`.  Regular API callers cannot
+override `user_id`; the server always uses their JWT identity.
+
 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

docs/settings.md

@@ -51,7 +51,7 @@ connection, Redis, auth secrets, server bind address). These stay in
 | Bridge identity | `[bridge]` | config.toml / env |
 | Console bind | `[console]` | config.toml / env |
 
-**ConfigStore settings** (~40 settings) are loaded from the database after
+**ConfigStore settings** (48 settings) are loaded from the database after
 storage initialization:
 
 | Section | Settings |
@@ -60,10 +60,12 @@ storage initialization:
 | `session` | instructions, retention_days, compact_max_tokens, auto_compact_pct |
 | `tools` | timeout, truncation, agent_max_turns, skip_permissions, search, search_threshold, search_max_results |
 | `server` | workstream_idle_timeout, max_workstreams |
+| `cluster` | node_fan_out_limit, mcp_max_servers |
 | `mcp` | config_path, refresh_interval, registry_url |
-| `ratelimit` | enabled, requests_per_second, burst |
+| `ratelimit` | enabled, requests_per_second, burst, trusted_proxies |
 | `health` | backend_probe_interval, backend_probe_timeout, circuit_breaker_threshold, circuit_breaker_cooldown |
-| `judge` | enabled, model, provider, base_url, api_key, confidence_threshold, max_context_ratio, timeout, read_only_tools |
+| `judge` | enabled, model, provider, base_url, api_key, confidence_threshold, max_context_ratio, timeout, read_only_tools, output_guard, redact_secrets |
+| `skills` | discovery_url |
 | `memory` | relevance_k, fetch_limit, max_content, nudge_cooldown, nudges |
 
 Settings are addressed by dotted key (e.g. `memory.relevance_k`). Each has a

docs/tls.md

@@ -0,0 +1,221 @@
+# TLS / mTLS
+
+Turnstone supports end-to-end transport encryption with mutual TLS (mTLS) for
+inter-service communication, powered by [lacme](https://pypi.org/project/lacme/).
+
+---
+
+## Quick Start (Docker Compose)
+
+```bash
+docker compose -f compose.yaml -f deploy/docker-compose.tls.yml up
+```
+
+This:
+1. Bootstraps an internal CA and issues certs for Redis/PostgreSQL
+2. Starts the console with TLS enabled (internal CA + ACME server)
+3. Server nodes auto-provision certs via the console's ACME endpoint
+4. All inter-service communication uses mTLS
+
+---
+
+## Architecture
+
+```
+Console (CA + ACME Server)
+  +-- CertificateAuthority (owns root key, signs certs)
+  +-- ACMEResponder (mounted at /acme, RFC 8555)
+  +-- GET /acme/ca.pem (root cert for node bootstrapping)
+                  |
+                  | ACME protocol (auto-approve, no challenge validation)
+      +-----------+-----------+
+      |           |           |
+  Server(s)    Bridge    Channel GW
+  (auto-cert    (mTLS     (mTLS
+   + renewal)   client)    client)
+```
+
+**Two cert paths on the console:**
+- **Internal cert** (mTLS): Always from the internal CA. Used for cluster
+  service mesh communication.
+- **Frontend cert** (HTTPS): From an external ACME CA (e.g. Let's Encrypt)
+  if `tls.acme_directory` is set, otherwise self-issued from the internal CA.
+
+---
+
+## Configuration
+
+### Settings (ConfigStore / Admin Settings tab)
+
+| Setting | Default | Description |
+|---------|---------|-------------|
+| `tls.enabled` | `false` | Master switch for internal mTLS |
+| `tls.acme_directory` | `""` | External ACME CA URL for console frontend cert |
+
+### Bootstrap Config (config.toml)
+
+These are needed before storage is available:
+
+```toml
+[redis]
+tls = false
+tls_ca = ""      # path to CA cert
+tls_cert = ""    # path to client cert
+tls_key = ""     # path to client key
+
+[database]
+sslmode = "prefer"   # disable, allow, prefer, require, verify-full
+sslrootcert = ""     # path to CA cert
+sslcert = ""         # path to client cert
+sslkey = ""          # path to client key
+```
+
+### Hardcoded Defaults
+
+| Parameter | Value | Notes |
+|-----------|-------|-------|
+| CA common name | "Turnstone CA" | |
+| CA validity | 10 years | |
+| Cert validity | 48 hours | Short-lived, auto-renewed |
+| Renewal interval | 24 hours | Half of validity |
+| ACME auto-approve | true | Internal network, no challenge validation |
+
+---
+
+## CLI
+
+### Offline Bootstrap
+
+Create a CA and infrastructure certs without a running console:
+
+```bash
+# Bootstrap CA + Redis + PostgreSQL certs
+turnstone-admin tls-bootstrap --out /certs --issue redis --issue postgres
+
+# Output:
+#   /certs/ca.pem              (CA root certificate)
+#   /certs/certs/redis/        (Redis cert + key)
+#   /certs/certs/postgres/     (PostgreSQL cert + key)
+```
+
+The output directory is chmod 0700 (contains the CA private key).
+
+### Online Cert Issuance
+
+Request certs from a running console's ACME endpoint:
+
+```bash
+# Download CA root cert (TOFU — verify fingerprint)
+turnstone-admin tls-ca-cert --out ca.pem --console-url http://console:8080
+
+# Request a cert for a domain
+turnstone-admin tls-issue worker-1.internal --out /certs --console-url http://console:8080
+
+# List issued certs
+turnstone-admin tls-list --console-url http://console:8080 --auth-token $TOKEN
+```
+
+### Console URL Discovery
+
+If `--console-url` is not provided, the CLI discovers it from the `services`
+table in the shared database. The console registers itself on startup.
+
+---
+
+## Admin UI
+
+The **TLS** tab in the console admin panel (System group) shows:
+- CA status (common name, certificate count)
+- Certificate table (domain, SANs, issued, expires)
+- Force-renew and delete actions per certificate
+
+---
+
+## SDK
+
+### Python
+
+```python
+from turnstone.sdk import TurnstoneServer
+
+client = TurnstoneServer(
+    base_url="https://server:8080",
+    token="tok_xxx",
+    ca_cert="/path/to/ca.pem",
+    client_cert="/path/to/cert.pem",
+    client_key="/path/to/key.pem",
+)
+```
+
+### TypeScript
+
+```typescript
+import { TurnstoneServer } from "@turnstone/sdk";
+import { Agent } from "undici";
+import * as fs from "fs";
+
+const agent = new Agent({
+  connect: {
+    ca: fs.readFileSync("/path/to/ca.pem"),
+    cert: fs.readFileSync("/path/to/cert.pem"),
+    key: fs.readFileSync("/path/to/key.pem"),
+  },
+});
+
+const client = new TurnstoneServer({
+  baseUrl: "https://server:8080",
+  token: "tok_xxx",
+  // Node.js 18+ uses undici under the hood
+  fetch: (url, init) =>
+    fetch(url, { ...init, dispatcher: agent } as RequestInit),
+});
+```
+
+---
+
+## How It Works
+
+### Node Bootstrap Flow
+
+1. Node starts, connects to shared database (plain connection)
+2. Discovers console URL from `services` table
+3. Fetches CA root cert from `http://console/acme/ca.pem` (plain HTTP, TOFU)
+4. Requests service cert via ACME protocol (plain HTTP, JWS-signed)
+5. Starts auto-renewal (24h interval, re-issues before expiry)
+6. All subsequent inter-service communication uses mTLS
+
+### Console Startup Flow
+
+1. Read `tls.enabled` from ConfigStore
+2. Initialize CA (load from DB or generate new root key)
+3. Mount ACME responder at `/acme` (serves `/ca.pem` natively)
+4. Issue console certs (internal + optional frontend)
+5. Start CA-direct auto-renewal (no network, signs directly)
+6. Register console URL in services table with heartbeat
+
+---
+
+## Troubleshooting
+
+### Cert expired / mTLS connection refused
+
+Certs are valid for 48 hours. If auto-renewal stopped (e.g. console was down),
+restart the service to re-request a cert.
+
+### "No console service found"
+
+The console registers itself in the `services` table on startup. If the console
+hasn't started or the registration expired (1 hour TTL), nodes can't discover
+it. Use `--console-url` explicitly.
+
+### Let's Encrypt for console frontend
+
+Set `tls.acme_directory` to `https://acme-v02.api.letsencrypt.org/directory`
+in the admin Settings tab. The console will request a publicly trusted cert
+for its HTTPS endpoint. Internal mTLS still uses the private CA.
+
+### Verifying the cert chain
+
+```bash
+openssl s_client -connect server:8080 -CAfile ca.pem
+```

docs/tools.md

@@ -492,7 +492,7 @@ data.get("mergedAt") is not None
 
 ---
 
-### load_skill
+### skill
 
 Discover and activate skills at runtime during a conversation. The model can
 search for available skills and load one by name, replacing the current active
@@ -543,7 +543,7 @@ pre-configure skills at workstream creation.
 | `watch`      | Monitor    | No (create)  | No    | No         | `command`   |
 | `read_resource`| MCP      | No           | Yes   | Yes        | `uri`       |
 | `use_prompt` | MCP        | No           | Yes   | Yes        | `name`      |
-| `load_skill` | Skills     | No (load)    | No    | No         | `name`      |
+| `skill`      | Skills     | No (load)    | No    | No         | `name`      |
 | `tool_search`| Search     | Yes          | No    | No         | `query`     |
 
 ---
@@ -591,9 +591,9 @@ CLI flags override the config file:
 
 ### How it works
 
-1. **Threshold check**: At session startup, `ToolSearchManager.should_activate()`
-   counts total tools (built-in + MCP). If the count is below the threshold, tool
-   search stays off and all tools are sent to the model directly.
+1. **Threshold check**: At session startup, if the total tool count (built-in + MCP)
+   is below the threshold, tool search stays off and all tools are sent to the model
+   directly.
 
 2. **Partitioning**: When active, tools are split into two sets:
    - **Always-on** -- the 17 built-in tools (members of `BUILTIN_TOOL_NAMES`).

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "turnstone"
-version = "0.8.2"
+version = "0.8.9"
 description = "Multi-node AI orchestration platform with tool use, agent routing, and cluster simulation."
 readme = "README.md"
 license = "BUSL-1.1"
@@ -51,8 +51,10 @@ console = ["redis>=7.2", "croniter>=3.0"]
 sim = ["redis>=7.2"]
 anthropic = ["anthropic>=0.39"]
 postgres = ["psycopg[binary]>=3.2"]
+ddg = ["ddgs>=9.0"]
 discord = ["discord.py>=2.4", "redis>=7.2"]
-all = ["turnstone[mq,console,sim,anthropic,postgres,discord]"]
+tls = ["lacme>=1.0.4"]
+all = ["turnstone[mq,console,sim,anthropic,postgres,discord,ddg,tls]"]
 
 [project.scripts]
 turnstone = "turnstone.cli:main"
@@ -77,7 +79,7 @@ include = [
   "turnstone/console/static/*.js",
   "turnstone/shared_static/*.css",
   "turnstone/shared_static/*.js",
-  "turnstone/shared_static/katex-0.16.38/**/*",
+  "turnstone/shared_static/katex-0.16.42/**/*",
   "turnstone/shared_static/hljs-11.11.1/**/*",
   "turnstone/shared_static/mermaid-11.13.0/**/*",
   "turnstone/sdk/py.typed",
@@ -164,6 +166,14 @@ ignore_missing_imports = true
 module = ["frontmatter", "frontmatter.*"]
 ignore_missing_imports = true
 
+[[tool.mypy.overrides]]
+module = ["ddgs", "ddgs.*"]
+ignore_missing_imports = true
+
+[[tool.mypy.overrides]]
+module = ["lacme", "lacme.*"]
+ignore_missing_imports = true
+
 [[tool.mypy.overrides]]
 module = ["turnstone.channels.discord.*"]
 disallow_subclassing_any = false

scripts/update-vendored-js.sh

@@ -28,9 +28,19 @@ usage() {
 LIB="$1"
 VERSION="$2"
 
-# Detect current version from pyproject.toml
+# Detect current version from the filesystem (not pyproject.toml, which
+# Renovate may have already updated).  Falls back to pyproject.toml if
+# no directory is found.
 detect_old_version() {
     local pattern="$1"
+    # Look for existing directory: e.g. turnstone/shared_static/katex-0.16.38
+    local dir
+    dir=$(find "${STATIC_DIR}" -maxdepth 1 -type d -name "${pattern}-*" | head -1)
+    if [[ -n "$dir" ]]; then
+        basename "$dir" | sed "s/${pattern}-//"
+        return
+    fi
+    # Fallback to pyproject.toml
     grep -oE "${pattern}-[0-9.]+" pyproject.toml | head -1 | sed "s/${pattern}-//"
 }
 
@@ -51,9 +61,19 @@ update_refs() {
     done
 }
 
+check_same_version() {
+    if [[ "$1" == "$2" ]]; then
+        echo "ERROR: Old version ($1) == new version ($2). Nothing to update."
+        echo "If the old directory was already removed, re-download with:"
+        echo "  rm -rf ${STATIC_DIR}/${3}-${1} && $0 $3 $2"
+        exit 1
+    fi
+}
+
 case "$LIB" in
     katex)
         OLD_VERSION=$(detect_old_version "katex")
+        check_same_version "$OLD_VERSION" "$VERSION" "katex"
         OLD_DIR="${STATIC_DIR}/katex-${OLD_VERSION}"
         NEW_DIR="${STATIC_DIR}/katex-${VERSION}"
 
@@ -87,6 +107,7 @@ case "$LIB" in
 
     hljs)
         OLD_VERSION=$(detect_old_version "hljs")
+        check_same_version "$OLD_VERSION" "$VERSION" "hljs"
         OLD_DIR="${STATIC_DIR}/hljs-${OLD_VERSION}"
         NEW_DIR="${STATIC_DIR}/hljs-${VERSION}"
 
@@ -107,6 +128,7 @@ case "$LIB" in
 
     mermaid)
         OLD_VERSION=$(detect_old_version "mermaid")
+        check_same_version "$OLD_VERSION" "$VERSION" "mermaid"
         OLD_DIR="${STATIC_DIR}/mermaid-${OLD_VERSION}"
         NEW_DIR="${STATIC_DIR}/mermaid-${VERSION}"
 

sdk/typescript/openapi-console.json

@@ -2,7 +2,7 @@
   "openapi": "3.1.0",
   "info": {
     "title": "turnstone Console API",
-    "version": "0.7.0",
+    "version": "0.8.4",
     "description": "Cluster-wide visibility and control across all turnstone nodes."
   },
   "paths": {
@@ -4378,6 +4378,12 @@
             "description": "Skill name (replaces default skills)",
             "title": "Skill",
             "type": "string"
+          },
+          "resume_ws": {
+            "default": "",
+            "description": "Workstream ID to resume (loads previous conversation)",
+            "title": "Resume Ws",
+            "type": "string"
           }
         },
         "title": "ConsoleCreateWsRequest",
@@ -6878,11 +6884,26 @@
             "title": "Enabled",
             "type": "boolean"
           },
+          "priority": {
+            "default": 0,
+            "title": "Priority",
+            "type": "integer"
+          },
           "allowed_tools": {
             "default": "[]",
             "title": "Allowed Tools",
             "type": "string"
           },
+          "license": {
+            "default": "",
+            "title": "License",
+            "type": "string"
+          },
+          "compatibility": {
+            "default": "",
+            "title": "Compatibility",
+            "type": "string"
+          },
           "scan_status": {
             "default": "",
             "title": "Scan Status",
@@ -7103,10 +7124,25 @@
             "title": "Enabled",
             "type": "boolean"
           },
+          "priority": {
+            "default": 0,
+            "title": "Priority",
+            "type": "integer"
+          },
           "allowed_tools": {
             "default": "[]",
             "title": "Allowed Tools",
             "type": "string"
+          },
+          "license": {
+            "default": "",
+            "title": "License",
+            "type": "string"
+          },
+          "compatibility": {
+            "default": "",
+            "title": "Compatibility",
+            "type": "string"
           }
         },
         "required": [
@@ -7346,6 +7382,18 @@
             "default": null,
             "title": "Enabled"
           },
+          "priority": {
+            "anyOf": [
+              {
+                "type": "integer"
+              },
+              {
+                "type": "null"
+              }
+            ],
+            "default": null,
+            "title": "Priority"
+          },
           "allowed_tools": {
             "anyOf": [
               {
@@ -7357,6 +7405,30 @@
             ],
             "default": null,
             "title": "Allowed Tools"
+          },
+          "license": {
+            "anyOf": [
+              {
+                "type": "string"
+              },
+              {
+                "type": "null"
+              }
+            ],
+            "default": null,
+            "title": "License"
+          },
+          "compatibility": {
+            "anyOf": [
+              {
+                "type": "string"
+              },
+              {
+                "type": "null"
+              }
+            ],
+            "default": null,
+            "title": "Compatibility"
           }
         },
         "title": "UpdateSkillRequest",

sdk/typescript/openapi-server.json

@@ -2,7 +2,7 @@
   "openapi": "3.1.0",
   "info": {
     "title": "turnstone Server API",
-    "version": "0.7.0",
+    "version": "0.8.4",
     "description": "Single-node workstream management, chat interaction, and real-time streaming."
   },
   "paths": {
@@ -1559,6 +1559,11 @@
             "title": "Version",
             "type": "string"
           },
+          "node_id": {
+            "default": "",
+            "title": "Node Id",
+            "type": "string"
+          },
           "uptime_seconds": {
             "default": 0.0,
             "title": "Uptime Seconds",
@@ -1569,6 +1574,12 @@
             "title": "Model",
             "type": "string"
           },
+          "max_ws": {
+            "default": 10,
+            "description": "Maximum concurrent workstreams",
+            "title": "Max Ws",
+            "type": "integer"
+          },
           "workstreams": {
             "$ref": "#/components/schemas/WorkstreamCounts",
             "default": {

sdk/typescript/package-lock.json

@@ -9,14 +9,14 @@
       "version": "0.3.0",
       "license": "BUSL-1.1",
       "devDependencies": {
-        "typescript": "^5.4",
+        "typescript": "^6.0.0",
         "vitest": "^4.1"
       }
     },
     "node_modules/@emnapi/core": {
-      "version": "1.9.0",
-      "resolved": "https://registry.npmjs.org/@emnapi/core/-/core-1.9.0.tgz",
-      "integrity": "sha512-0DQ98G9ZQZOxfUcQn1waV2yS8aWdZ6kJMbYCJB3oUBecjWYO1fqJ+a1DRfPF3O5JEkwqwP1A9QEN/9mYm2Yd0w==",
+      "version": "1.9.1",
+      "resolved": "https://registry.npmjs.org/@emnapi/core/-/core-1.9.1.tgz",
+      "integrity": "sha512-mukuNALVsoix/w1BJwFzwXBN/dHeejQtuVzcDsfOEsdpCumXb/E9j8w11h5S54tT1xhifGfbbSm/ICrObRb3KA==",
       "dev": true,
       "license": "MIT",
       "optional": true,
@@ -26,9 +26,9 @@
       }
     },
     "node_modules/@emnapi/runtime": {
-      "version": "1.9.0",
-      "resolved": "https://registry.npmjs.org/@emnapi/runtime/-/runtime-1.9.0.tgz",
-      "integrity": "sha512-QN75eB0IH2ywSpRpNddCRfQIhmJYBCJ1x5Lb3IscKAL8bMnVAKnRg8dCoXbHzVLLH7P38N2Z3mtulB7W0J0FKw==",
+      "version": "1.9.1",
+      "resolved": "https://registry.npmjs.org/@emnapi/runtime/-/runtime-1.9.1.tgz",
+      "integrity": "sha512-VYi5+ZVLhpgK4hQ0TAjiQiZ6ol0oe4mBx7mVv7IflsiEp0OWoVsp/+f9Vc1hOhE0TtkORVrI1GvzyreqpgWtkA==",
       "dev": true,
       "license": "MIT",
       "optional": true,
@@ -71,20 +71,10 @@
         "url": "https://github.com/sponsors/Brooooooklyn"
       }
     },
-    "node_modules/@oxc-project/runtime": {
-      "version": "0.115.0",
-      "resolved": "https://registry.npmjs.org/@oxc-project/runtime/-/runtime-0.115.0.tgz",
-      "integrity": "sha512-Rg8Wlt5dCbXhQnsXPrkOjL1DTSvXLgb2R/KYfnf1/K+R0k6UMLEmbQXPM+kwrWqSmWA2t0B1EtHy2/3zikQpvQ==",
-      "dev": true,
-      "license": "MIT",
-      "engines": {
-        "node": "^20.19.0 || >=22.12.0"
-      }
-    },
     "node_modules/@oxc-project/types": {
-      "version": "0.115.0",
-      "resolved": "https://registry.npmjs.org/@oxc-project/types/-/types-0.115.0.tgz",
-      "integrity": "sha512-4n91DKnebUS4yjUHl2g3/b2T+IUdCfmoZGhmwsovZCDaJSs+QkVAM+0AqqTxHSsHfeiMuueT75cZaZcT/m0pSw==",
+      "version": "0.120.0",
+      "resolved": "https://registry.npmjs.org/@oxc-project/types/-/types-0.120.0.tgz",
+      "integrity": "sha512-k1YNu55DuvAip/MGE1FTsIuU3FUCn6v/ujG9V7Nq5Df/kX2CWb13hhwD0lmJGMGqE+bE1MXvv9SZVnMzEXlWcg==",
       "dev": true,
       "license": "MIT",
       "funding": {
@@ -92,9 +82,9 @@
       }
     },
     "node_modules/@rolldown/binding-android-arm64": {
-      "version": "1.0.0-rc.9",
-      "resolved": "https://registry.npmjs.org/@rolldown/binding-android-arm64/-/binding-android-arm64-1.0.0-rc.9.tgz",
-      "integrity": "sha512-lcJL0bN5hpgJfSIz/8PIf02irmyL43P+j1pTCfbD1DbLkmGRuFIA4DD3B3ZOvGqG0XiVvRznbKtN0COQVaKUTg==",
+      "version": "1.0.0-rc.10",
+      "resolved": "https://registry.npmjs.org/@rolldown/binding-android-arm64/-/binding-android-arm64-1.0.0-rc.10.tgz",
+      "integrity": "sha512-jOHxwXhxmFKuXztiu1ORieJeTbx5vrTkcOkkkn2d35726+iwhrY1w/+nYY/AGgF12thg33qC3R1LMBF5tHTZHg==",
       "cpu": [
         "arm64"
       ],
@@ -109,9 +99,9 @@
       }
     },
     "node_modules/@rolldown/binding-darwin-arm64": {
-      "version": "1.0.0-rc.9",
-      "resolved": "https://registry.npmjs.org/@rolldown/binding-darwin-arm64/-/binding-darwin-arm64-1.0.0-rc.9.tgz",
-      "integrity": "sha512-J7Zk3kLYFsLtuH6U+F4pS2sYVzac0qkjcO5QxHS7OS7yZu2LRs+IXo+uvJ/mvpyUljDJ3LROZPoQfgBIpCMhdQ==",
+      "version": "1.0.0-rc.10",
+      "resolved": "https://registry.npmjs.org/@rolldown/binding-darwin-arm64/-/binding-darwin-arm64-1.0.0-rc.10.tgz",
+      "integrity": "sha512-gED05Teg/vtTZbIJBc4VNMAxAFDUPkuO/rAIyyxZjTj1a1/s6z5TII/5yMGZ0uLRCifEtwUQn8OlYzuYc0m70w==",
       "cpu": [
         "arm64"
       ],
@@ -126,9 +116,9 @@
       }
     },
     "node_modules/@rolldown/binding-darwin-x64": {
-      "version": "1.0.0-rc.9",
-      "resolved": "https://registry.npmjs.org/@rolldown/binding-darwin-x64/-/binding-darwin-x64-1.0.0-rc.9.tgz",
-      "integrity": "sha512-iwtmmghy8nhfRGeNAIltcNXzD0QMNaaA5U/NyZc1Ia4bxrzFByNMDoppoC+hl7cDiUq5/1CnFthpT9n+UtfFyg==",
+      "version": "1.0.0-rc.10",
+      "resolved": "https://registry.npmjs.org/@rolldown/binding-darwin-x64/-/binding-darwin-x64-1.0.0-rc.10.tgz",
+      "integrity": "sha512-rI15NcM1mA48lqrIxVkHfAqcyFLcQwyXWThy+BQ5+mkKKPvSO26ir+ZDp36AgYoYVkqvMcdS8zOE6SeBsR9e8A==",
       "cpu": [
         "x64"
       ],
@@ -143,9 +133,9 @@
       }
     },
     "node_modules/@rolldown/binding-freebsd-x64": {
-      "version": "1.0.0-rc.9",
-      "resolved": "https://registry.npmjs.org/@rolldown/binding-freebsd-x64/-/binding-freebsd-x64-1.0.0-rc.9.tgz",
-      "integrity": "sha512-DLFYI78SCiZr5VvdEplsVC2Vx53lnA4/Ga5C65iyldMVaErr86aiqCoNBLl92PXPfDtUYjUh+xFFor40ueNs4Q==",
+      "version": "1.0.0-rc.10",
+      "resolved": "https://registry.npmjs.org/@rolldown/binding-freebsd-x64/-/binding-freebsd-x64-1.0.0-rc.10.tgz",
+      "integrity": "sha512-XZRXHdTa+4ME1MuDVp021+doQ+z6Ei4CCFmNc5/sKbqb8YmkiJdj8QKlV3rCI0AJtAeSB5n0WGPuJWNL9p/L2w==",
       "cpu": [
         "x64"
       ],
@@ -160,9 +150,9 @@
       }
     },
     "node_modules/@rolldown/binding-linux-arm-gnueabihf": {
-      "version": "1.0.0-rc.9",
-      "resolved": "https://registry.npmjs.org/@rolldown/binding-linux-arm-gnueabihf/-/binding-linux-arm-gnueabihf-1.0.0-rc.9.tgz",
-      "integrity": "sha512-CsjTmTwd0Hri6iTw/DRMK7kOZ7FwAkrO4h8YWKoX/kcj833e4coqo2wzIFywtch/8Eb5enQ/lwLM7w6JX1W5RQ==",
+      "version": "1.0.0-rc.10",
+      "resolved": "https://registry.npmjs.org/@rolldown/binding-linux-arm-gnueabihf/-/binding-linux-arm-gnueabihf-1.0.0-rc.10.tgz",
+      "integrity": "sha512-R0SQMRluISSLzFE20sPWYHVmJdDQnRyc/FzSCN72BqQmh2SOZUFG+N3/vBZpR4C6WpEUVYJLrYUXaj43sJsNLA==",
       "cpu": [
         "arm"
       ],
@@ -177,9 +167,9 @@
       }
     },
     "node_modules/@rolldown/binding-linux-arm64-gnu": {
-      "version": "1.0.0-rc.9",
-      "resolved": "https://registry.npmjs.org/@rolldown/binding-linux-arm64-gnu/-/binding-linux-arm64-gnu-1.0.0-rc.9.tgz",
-      "integrity": "sha512-2x9O2JbSPxpxMDhP9Z74mahAStibTlrBMW0520+epJH5sac7/LwZW5Bmg/E6CXuEF53JJFW509uP+lSedaUNxg==",
+      "version": "1.0.0-rc.10",
+      "resolved": "https://registry.npmjs.org/@rolldown/binding-linux-arm64-gnu/-/binding-linux-arm64-gnu-1.0.0-rc.10.tgz",
+      "integrity": "sha512-Y1reMrV/o+cwpduYhJuOE3OMKx32RMYCidf14y+HssARRmhDuWXJ4yVguDg2R/8SyyGNo+auzz64LnPK9Hq6jg==",
       "cpu": [
         "arm64"
       ],
@@ -194,9 +184,9 @@
       }
     },
     "node_modules/@rolldown/binding-linux-arm64-musl": {
-      "version": "1.0.0-rc.9",
-      "resolved": "https://registry.npmjs.org/@rolldown/binding-linux-arm64-musl/-/binding-linux-arm64-musl-1.0.0-rc.9.tgz",
-      "integrity": "sha512-JA1QRW31ogheAIRhIg9tjMfsYbglXXYGNPLdPEYrwFxdbkQCAzvpSCSHCDWNl4hTtrol8WeboCSEpjdZK8qrCg==",
+      "version": "1.0.0-rc.10",
+      "resolved": "https://registry.npmjs.org/@rolldown/binding-linux-arm64-musl/-/binding-linux-arm64-musl-1.0.0-rc.10.tgz",
+      "integrity": "sha512-vELN+HNb2IzuzSBUOD4NHmP9yrGwl1DVM29wlQvx1OLSclL0NgVWnVDKl/8tEks79EFek/kebQKnNJkIAA4W2g==",
       "cpu": [
         "arm64"
       ],
@@ -211,9 +201,9 @@
       }
     },
     "node_modules/@rolldown/binding-linux-ppc64-gnu": {
-      "version": "1.0.0-rc.9",
-      "resolved": "https://registry.npmjs.org/@rolldown/binding-linux-ppc64-gnu/-/binding-linux-ppc64-gnu-1.0.0-rc.9.tgz",
-      "integrity": "sha512-aOKU9dJheda8Kj8Y3w9gnt9QFOO+qKPAl8SWd7JPHP+Cu0EuDAE5wokQubLzIDQWg2myXq2XhTpOVS07qqvT+w==",
+      "version": "1.0.0-rc.10",
+      "resolved": "https://registry.npmjs.org/@rolldown/binding-linux-ppc64-gnu/-/binding-linux-ppc64-gnu-1.0.0-rc.10.tgz",
+      "integrity": "sha512-ZqrufYTgzxbHwpqOjzSsb0UV/aV2TFIY5rP8HdsiPTv/CuAgCRjM6s9cYFwQ4CNH+hf9Y4erHW1GjZuZ7WoI7w==",
       "cpu": [
         "ppc64"
       ],
@@ -228,9 +218,9 @@
       }
     },
     "node_modules/@rolldown/binding-linux-s390x-gnu": {
-      "version": "1.0.0-rc.9",
-      "resolved": "https://registry.npmjs.org/@rolldown/binding-linux-s390x-gnu/-/binding-linux-s390x-gnu-1.0.0-rc.9.tgz",
-      "integrity": "sha512-OalO94fqj7IWRn3VdXWty75jC5dk4C197AWEuMhIpvVv2lw9fiPhud0+bW2ctCxb3YoBZor71QHbY+9/WToadA==",
+      "version": "1.0.0-rc.10",
+      "resolved": "https://registry.npmjs.org/@rolldown/binding-linux-s390x-gnu/-/binding-linux-s390x-gnu-1.0.0-rc.10.tgz",
+      "integrity": "sha512-gSlmVS1FZJSRicA6IyjoRoKAFK7IIHBs7xJuHRSmjImqk3mPPWbR7RhbnfH2G6bcmMEllCt2vQ/7u9e6bBnByg==",
       "cpu": [
         "s390x"
       ],
@@ -245,9 +235,9 @@
       }
     },
     "node_modules/@rolldown/binding-linux-x64-gnu": {
-      "version": "1.0.0-rc.9",
-      "resolved": "https://registry.npmjs.org/@rolldown/binding-linux-x64-gnu/-/binding-linux-x64-gnu-1.0.0-rc.9.tgz",
-      "integrity": "sha512-cVEl1vZtBsBZna3YMjGXNvnYYrOJ7RzuWvZU0ffvJUexWkukMaDuGhUXn0rjnV0ptzGVkvc+vW9Yqy6h8YX4pg==",
+      "version": "1.0.0-rc.10",
+      "resolved": "https://registry.npmjs.org/@rolldown/binding-linux-x64-gnu/-/binding-linux-x64-gnu-1.0.0-rc.10.tgz",
+      "integrity": "sha512-eOCKUpluKgfObT2pHjztnaWEIbUabWzk3qPZ5PuacuPmr4+JtQG4k2vGTY0H15edaTnicgU428XW/IH6AimcQw==",
       "cpu": [
         "x64"
       ],
@@ -262,9 +252,9 @@
       }
     },
     "node_modules/@rolldown/binding-linux-x64-musl": {
-      "version": "1.0.0-rc.9",
-      "resolved": "https://registry.npmjs.org/@rolldown/binding-linux-x64-musl/-/binding-linux-x64-musl-1.0.0-rc.9.tgz",
-      "integrity": "sha512-UzYnKCIIc4heAKgI4PZ3dfBGUZefGCJ1TPDuLHoCzgrMYPb5Rv6TLFuYtyM4rWyHM7hymNdsg5ik2C+UD9VDbA==",
+      "version": "1.0.0-rc.10",
+      "resolved": "https://registry.npmjs.org/@rolldown/binding-linux-x64-musl/-/binding-linux-x64-musl-1.0.0-rc.10.tgz",
+      "integrity": "sha512-Xdf2jQbfQowJnLcgYfD/m0Uu0Qj5OdxKallD78/IPPfzaiaI4KRAwZzHcKQ4ig1gtg1SuzC7jovNiM2TzQsBXA==",
       "cpu": [
         "x64"
       ],
@@ -279,9 +269,9 @@
       }
     },
     "node_modules/@rolldown/binding-openharmony-arm64": {
-      "version": "1.0.0-rc.9",
-      "resolved": "https://registry.npmjs.org/@rolldown/binding-openharmony-arm64/-/binding-openharmony-arm64-1.0.0-rc.9.tgz",
-      "integrity": "sha512-+6zoiF+RRyf5cdlFQP7nm58mq7+/2PFaY2DNQeD4B87N36JzfF/l9mdBkkmTvSYcYPE8tMh/o3cRlsx1ldLfog==",
+      "version": "1.0.0-rc.10",
+      "resolved": "https://registry.npmjs.org/@rolldown/binding-openharmony-arm64/-/binding-openharmony-arm64-1.0.0-rc.10.tgz",
+      "integrity": "sha512-o1hYe8hLi1EY6jgPFyxQgQ1wcycX+qz8eEbVmot2hFkgUzPxy9+kF0u0NIQBeDq+Mko47AkaFFaChcvZa9UX9Q==",
       "cpu": [
         "arm64"
       ],
@@ -296,9 +286,9 @@
       }
     },
     "node_modules/@rolldown/binding-wasm32-wasi": {
-      "version": "1.0.0-rc.9",
-      "resolved": "https://registry.npmjs.org/@rolldown/binding-wasm32-wasi/-/binding-wasm32-wasi-1.0.0-rc.9.tgz",
-      "integrity": "sha512-rgFN6sA/dyebil3YTlL2evvi/M+ivhfnyxec7AccTpRPccno/rPoNlqybEZQBkcbZu8Hy+eqNJCqfBR8P7Pg8g==",
+      "version": "1.0.0-rc.10",
+      "resolved": "https://registry.npmjs.org/@rolldown/binding-wasm32-wasi/-/binding-wasm32-wasi-1.0.0-rc.10.tgz",
+      "integrity": "sha512-Ugv9o7qYJudqQO5Y5y2N2SOo6S4WiqiNOpuQyoPInnhVzCY+wi/GHltcLHypG9DEUYMB0iTB/huJrpadiAcNcA==",
       "cpu": [
         "wasm32"
       ],
@@ -313,9 +303,9 @@
       }
     },
     "node_modules/@rolldown/binding-win32-arm64-msvc": {
-      "version": "1.0.0-rc.9",
-      "resolved": "https://registry.npmjs.org/@rolldown/binding-win32-arm64-msvc/-/binding-win32-arm64-msvc-1.0.0-rc.9.tgz",
-      "integrity": "sha512-lHVNUG/8nlF1IQk1C0Ci574qKYyty2goMiPlRqkC5R+3LkXDkL5Dhx8ytbxq35m+pkHVIvIxviD+TWLdfeuadA==",
+      "version": "1.0.0-rc.10",
+      "resolved": "https://registry.npmjs.org/@rolldown/binding-win32-arm64-msvc/-/binding-win32-arm64-msvc-1.0.0-rc.10.tgz",
+      "integrity": "sha512-7UODQb4fQUNT/vmgDZBl3XOBAIOutP5R3O/rkxg0aLfEGQ4opbCgU5vOw/scPe4xOqBwL9fw7/RP1vAMZ6QlAQ==",
       "cpu": [
         "arm64"
       ],
@@ -330,9 +320,9 @@
       }
     },
     "node_modules/@rolldown/binding-win32-x64-msvc": {
-      "version": "1.0.0-rc.9",
-      "resolved": "https://registry.npmjs.org/@rolldown/binding-win32-x64-msvc/-/binding-win32-x64-msvc-1.0.0-rc.9.tgz",
-      "integrity": "sha512-G0oA4+w1iY5AGi5HcDTxWsoxF509hrFIPB2rduV5aDqS9FtDg1CAfa7V34qImbjfhIcA8C+RekocJZA96EarwQ==",
+      "version": "1.0.0-rc.10",
+      "resolved": "https://registry.npmjs.org/@rolldown/binding-win32-x64-msvc/-/binding-win32-x64-msvc-1.0.0-rc.10.tgz",
+      "integrity": "sha512-PYxKHMVHOb5NJuDL53vBUl1VwUjymDcYI6rzpIni0C9+9mTiJedvUxSk7/RPp7OOAm3v+EjgMu9bIy3N6b408w==",
       "cpu": [
         "x64"
       ],
@@ -347,9 +337,9 @@
       }
     },
     "node_modules/@rolldown/pluginutils": {
-      "version": "1.0.0-rc.9",
-      "resolved": "https://registry.npmjs.org/@rolldown/pluginutils/-/pluginutils-1.0.0-rc.9.tgz",
-      "integrity": "sha512-w6oiRWgEBl04QkFZgmW+jnU1EC9b57Oihi2ot3HNWIQRqgHp5PnYDia5iZ5FF7rpa4EQdiqMDXjlqKGXBhsoXw==",
+      "version": "1.0.0-rc.10",
+      "resolved": "https://registry.npmjs.org/@rolldown/pluginutils/-/pluginutils-1.0.0-rc.10.tgz",
+      "integrity": "sha512-UkVDEFk1w3mveXeKgaTuYfKWtPbvgck1dT8TUG3bnccrH0XtLTuAyfCoks4Q/M5ZGToSVJTIQYCzy2g/atAOeg==",
       "dev": true,
       "license": "MIT"
     },
@@ -397,16 +387,16 @@
       "license": "MIT"
     },
     "node_modules/@vitest/expect": {
-      "version": "4.1.0",
-      "resolved": "https://registry.npmjs.org/@vitest/expect/-/expect-4.1.0.tgz",
-      "integrity": "sha512-EIxG7k4wlWweuCLG9Y5InKFwpMEOyrMb6ZJ1ihYu02LVj/bzUwn2VMU+13PinsjRW75XnITeFrQBMH5+dLvCDA==",
+      "version": "4.1.1",
+      "resolved": "https://registry.npmjs.org/@vitest/expect/-/expect-4.1.1.tgz",
+      "integrity": "sha512-xAV0fqBTk44Rn6SjJReEQkHP3RrqbJo6JQ4zZ7/uVOiJZRarBtblzrOfFIZeYUrukp2YD6snZG6IBqhOoHTm+A==",
       "dev": true,
       "license": "MIT",
       "dependencies": {
         "@standard-schema/spec": "^1.1.0",
         "@types/chai": "^5.2.2",
-        "@vitest/spy": "4.1.0",
-        "@vitest/utils": "4.1.0",
+        "@vitest/spy": "4.1.1",
+        "@vitest/utils": "4.1.1",
         "chai": "^6.2.2",
         "tinyrainbow": "^3.0.3"
       },
@@ -415,13 +405,13 @@
       }
     },
     "node_modules/@vitest/mocker": {
-      "version": "4.1.0",
-      "resolved": "https://registry.npmjs.org/@vitest/mocker/-/mocker-4.1.0.tgz",
-      "integrity": "sha512-evxREh+Hork43+Y4IOhTo+h5lGmVRyjqI739Rz4RlUPqwrkFFDF6EMvOOYjTx4E8Tl6gyCLRL8Mu7Ry12a13Tw==",
+      "version": "4.1.1",
+      "resolved": "https://registry.npmjs.org/@vitest/mocker/-/mocker-4.1.1.tgz",
+      "integrity": "sha512-h3BOylsfsCLPeceuCPAAJ+BvNwSENgJa4hXoXu4im0bs9Lyp4URc4JYK4pWLZ4pG/UQn7AT92K6IByi6rE6g3A==",
       "dev": true,
       "license": "MIT",
       "dependencies": {
-        "@vitest/spy": "4.1.0",
+        "@vitest/spy": "4.1.1",
         "estree-walker": "^3.0.3",
         "magic-string": "^0.30.21"
       },
@@ -430,7 +420,7 @@
       },
       "peerDependencies": {
         "msw": "^2.4.9",
-        "vite": "^6.0.0 || ^7.0.0 || ^8.0.0-0"
+        "vite": "^6.0.0 || ^7.0.0 || ^8.0.0"
       },
       "peerDependenciesMeta": {
         "msw": {
@@ -442,9 +432,9 @@
       }
     },
     "node_modules/@vitest/pretty-format": {
-      "version": "4.1.0",
-      "resolved": "https://registry.npmjs.org/@vitest/pretty-format/-/pretty-format-4.1.0.tgz",
-      "integrity": "sha512-3RZLZlh88Ib0J7NQTRATfc/3ZPOnSUn2uDBUoGNn5T36+bALixmzphN26OUD3LRXWkJu4H0s5vvUeqBiw+kS0A==",
+      "version": "4.1.1",
+      "resolved": "https://registry.npmjs.org/@vitest/pretty-format/-/pretty-format-4.1.1.tgz",
+      "integrity": "sha512-GM+TEQN5WhOygr1lp7skeVjdLPqqWMHsfzXrcHAqZJi/lIVh63H0kaRCY8MDhNWikx19zBUK8ceaLB7X5AH9NQ==",
       "dev": true,
       "license": "MIT",
       "dependencies": {
@@ -455,13 +445,13 @@
       }
     },
     "node_modules/@vitest/runner": {
-      "version": "4.1.0",
-      "resolved": "https://registry.npmjs.org/@vitest/runner/-/runner-4.1.0.tgz",
-      "integrity": "sha512-Duvx2OzQ7d6OjchL+trw+aSrb9idh7pnNfxrklo14p3zmNL4qPCDeIJAK+eBKYjkIwG96Bc6vYuxhqDXQOWpoQ==",
+      "version": "4.1.1",
+      "resolved": "https://registry.npmjs.org/@vitest/runner/-/runner-4.1.1.tgz",
+      "integrity": "sha512-f7+FPy75vN91QGWsITueq0gedwUZy1fLtHOCMeQpjs8jTekAHeKP80zfDEnhrleviLHzVSDXIWuCIOFn3D3f8A==",
       "dev": true,
       "license": "MIT",
       "dependencies": {
-        "@vitest/utils": "4.1.0",
+        "@vitest/utils": "4.1.1",
         "pathe": "^2.0.3"
       },
       "funding": {
@@ -469,14 +459,14 @@
       }
     },
     "node_modules/@vitest/snapshot": {
-      "version": "4.1.0",
-      "resolved": "https://registry.npmjs.org/@vitest/snapshot/-/snapshot-4.1.0.tgz",
-      "integrity": "sha512-0Vy9euT1kgsnj1CHttwi9i9o+4rRLEaPRSOJ5gyv579GJkNpgJK+B4HSv/rAWixx2wdAFci1X4CEPjiu2bXIMg==",
+      "version": "4.1.1",
+      "resolved": "https://registry.npmjs.org/@vitest/snapshot/-/snapshot-4.1.1.tgz",
+      "integrity": "sha512-kMVSgcegWV2FibXEx9p9WIKgje58lcTbXgnJixfcg15iK8nzCXhmalL0ZLtTWLW9PH1+1NEDShiFFedB3tEgWg==",
       "dev": true,
       "license": "MIT",
       "dependencies": {
-        "@vitest/pretty-format": "4.1.0",
-        "@vitest/utils": "4.1.0",
+        "@vitest/pretty-format": "4.1.1",
+        "@vitest/utils": "4.1.1",
         "magic-string": "^0.30.21",
         "pathe": "^2.0.3"
       },
@@ -485,9 +475,9 @@
       }
     },
     "node_modules/@vitest/spy": {
-      "version": "4.1.0",
-      "resolved": "https://registry.npmjs.org/@vitest/spy/-/spy-4.1.0.tgz",
-      "integrity": "sha512-pz77k+PgNpyMDv2FV6qmk5ZVau6c3R8HC8v342T2xlFxQKTrSeYw9waIJG8KgV9fFwAtTu4ceRzMivPTH6wSxw==",
+      "version": "4.1.1",
+      "resolved": "https://registry.npmjs.org/@vitest/spy/-/spy-4.1.1.tgz",
+      "integrity": "sha512-6Ti/KT5OVaiupdIZEuZN7l3CZcR0cxnxt70Z0//3CtwgObwA6jZhmVBA3yrXSVN3gmwjgd7oDNLlsXz526gpRA==",
       "dev": true,
       "license": "MIT",
       "funding": {
@@ -495,13 +485,13 @@
       }
     },
     "node_modules/@vitest/utils": {
-      "version": "4.1.0",
-      "resolved": "https://registry.npmjs.org/@vitest/utils/-/utils-4.1.0.tgz",
-      "integrity": "sha512-XfPXT6a8TZY3dcGY8EdwsBulFCIw+BeeX0RZn2x/BtiY/75YGh8FeWGG8QISN/WhaqSrE2OrlDgtF8q5uhOTmw==",
+      "version": "4.1.1",
+      "resolved": "https://registry.npmjs.org/@vitest/utils/-/utils-4.1.1.tgz",
+      "integrity": "sha512-cNxAlaB3sHoCdL6pj6yyUXv9Gry1NHNg0kFTXdvSIZXLHsqKH7chiWOkwJ5s5+d/oMwcoG9T0bKU38JZWKusrQ==",
       "dev": true,
       "license": "MIT",
       "dependencies": {
-        "@vitest/pretty-format": "4.1.0",
+        "@vitest/pretty-format": "4.1.1",
         "convert-source-map": "^2.0.0",
         "tinyrainbow": "^3.0.3"
       },
@@ -922,11 +912,10 @@
       "license": "ISC"
     },
     "node_modules/picomatch": {
-      "version": "4.0.3",
-      "resolved": "https://registry.npmjs.org/picomatch/-/picomatch-4.0.3.tgz",
-      "integrity": "sha512-5gTmgEY/sqK6gFXLIsQNH19lWb4ebPDLA4SdLP7dsWkIXHWlG66oPuVvXSGFPppYZz8ZDZq0dYYrbHfBCVUb1Q==",
+      "version": "4.0.4",
+      "resolved": "https://registry.npmjs.org/picomatch/-/picomatch-4.0.4.tgz",
+      "integrity": "sha512-QP88BAKvMam/3NxH6vj2o21R6MjxZUAd6nlwAS/pnGvN9IVLocLHxGYIzFhg6fUQ+5th6P4dv4eW9jX3DSIj7A==",
       "dev": true,
-      "license": "MIT",
       "engines": {
         "node": ">=12"
       },
@@ -964,14 +953,14 @@
       }
     },
     "node_modules/rolldown": {
-      "version": "1.0.0-rc.9",
-      "resolved": "https://registry.npmjs.org/rolldown/-/rolldown-1.0.0-rc.9.tgz",
-      "integrity": "sha512-9EbgWge7ZH+yqb4d2EnELAntgPTWbfL8ajiTW+SyhJEC4qhBbkCKbqFV4Ge4zmu5ziQuVbWxb/XwLZ+RIO7E8Q==",
+      "version": "1.0.0-rc.10",
+      "resolved": "https://registry.npmjs.org/rolldown/-/rolldown-1.0.0-rc.10.tgz",
+      "integrity": "sha512-q7j6vvarRFmKpgJUT8HCAUljkgzEp4LAhPlJUvQhA5LA1SUL36s5QCysMutErzL3EbNOZOkoziSx9iZC4FddKA==",
       "dev": true,
       "license": "MIT",
       "dependencies": {
-        "@oxc-project/types": "=0.115.0",
-        "@rolldown/pluginutils": "1.0.0-rc.9"
+        "@oxc-project/types": "=0.120.0",
+        "@rolldown/pluginutils": "1.0.0-rc.10"
       },
       "bin": {
         "rolldown": "bin/cli.mjs"
@@ -980,21 +969,21 @@
         "node": "^20.19.0 || >=22.12.0"
       },
       "optionalDependencies": {
-        "@rolldown/binding-android-arm64": "1.0.0-rc.9",
-        "@rolldown/binding-darwin-arm64": "1.0.0-rc.9",
-        "@rolldown/binding-darwin-x64": "1.0.0-rc.9",
-        "@rolldown/binding-freebsd-x64": "1.0.0-rc.9",
-        "@rolldown/binding-linux-arm-gnueabihf": "1.0.0-rc.9",
-        "@rolldown/binding-linux-arm64-gnu": "1.0.0-rc.9",
-        "@rolldown/binding-linux-arm64-musl": "1.0.0-rc.9",
-        "@rolldown/binding-linux-ppc64-gnu": "1.0.0-rc.9",
-        "@rolldown/binding-linux-s390x-gnu": "1.0.0-rc.9",
-        "@rolldown/binding-linux-x64-gnu": "1.0.0-rc.9",
-        "@rolldown/binding-linux-x64-musl": "1.0.0-rc.9",
-        "@rolldown/binding-openharmony-arm64": "1.0.0-rc.9",
-        "@rolldown/binding-wasm32-wasi": "1.0.0-rc.9",
-        "@rolldown/binding-win32-arm64-msvc": "1.0.0-rc.9",
-        "@rolldown/binding-win32-x64-msvc": "1.0.0-rc.9"
+        "@rolldown/binding-android-arm64": "1.0.0-rc.10",
+        "@rolldown/binding-darwin-arm64": "1.0.0-rc.10",
+        "@rolldown/binding-darwin-x64": "1.0.0-rc.10",
+        "@rolldown/binding-freebsd-x64": "1.0.0-rc.10",
+        "@rolldown/binding-linux-arm-gnueabihf": "1.0.0-rc.10",
+        "@rolldown/binding-linux-arm64-gnu": "1.0.0-rc.10",
+        "@rolldown/binding-linux-arm64-musl": "1.0.0-rc.10",
+        "@rolldown/binding-linux-ppc64-gnu": "1.0.0-rc.10",
+        "@rolldown/binding-linux-s390x-gnu": "1.0.0-rc.10",
+        "@rolldown/binding-linux-x64-gnu": "1.0.0-rc.10",
+        "@rolldown/binding-linux-x64-musl": "1.0.0-rc.10",
+        "@rolldown/binding-openharmony-arm64": "1.0.0-rc.10",
+        "@rolldown/binding-wasm32-wasi": "1.0.0-rc.10",
+        "@rolldown/binding-win32-arm64-msvc": "1.0.0-rc.10",
+        "@rolldown/binding-win32-x64-msvc": "1.0.0-rc.10"
       }
     },
     "node_modules/siginfo": {
@@ -1081,9 +1070,9 @@
       "optional": true
     },
     "node_modules/typescript": {
-      "version": "5.9.3",
-      "resolved": "https://registry.npmjs.org/typescript/-/typescript-5.9.3.tgz",
-      "integrity": "sha512-jl1vZzPDinLr9eUt3J/t7V6FgNEw9QjvBPdysz9KfQDD41fQrC2Y4vKQdiaUpFT4bXlb1RHhLpp8wtm6M5TgSw==",
+      "version": "6.0.2",
+      "resolved": "https://registry.npmjs.org/typescript/-/typescript-6.0.2.tgz",
+      "integrity": "sha512-bGdAIrZ0wiGDo5l8c++HWtbaNCWTS4UTv7RaTH/ThVIgjkveJt83m74bBHMJkuCbslY8ixgLBVZJIOiQlQTjfQ==",
       "dev": true,
       "license": "Apache-2.0",
       "bin": {
@@ -1095,17 +1084,16 @@
       }
     },
     "node_modules/vite": {
-      "version": "8.0.0",
-      "resolved": "https://registry.npmjs.org/vite/-/vite-8.0.0.tgz",
-      "integrity": "sha512-fPGaRNj9Zytaf8LEiBhY7Z6ijnFKdzU/+mL8EFBaKr7Vw1/FWcTBAMW0wLPJAGMPX38ZPVCVgLceWiEqeoqL2Q==",
+      "version": "8.0.1",
+      "resolved": "https://registry.npmjs.org/vite/-/vite-8.0.1.tgz",
+      "integrity": "sha512-wt+Z2qIhfFt85uiyRt5LPU4oVEJBXj8hZNWKeqFG4gRG/0RaRGJ7njQCwzFVjO+v4+Ipmf5CY7VdmZRAYYBPHw==",
       "dev": true,
       "license": "MIT",
       "dependencies": {
-        "@oxc-project/runtime": "0.115.0",
         "lightningcss": "^1.32.0",
         "picomatch": "^4.0.3",
         "postcss": "^8.5.8",
-        "rolldown": "1.0.0-rc.9",
+        "rolldown": "1.0.0-rc.10",
         "tinyglobby": "^0.2.15"
       },
       "bin": {
@@ -1122,7 +1110,7 @@
       },
       "peerDependencies": {
         "@types/node": "^20.19.0 || >=22.12.0",
-        "@vitejs/devtools": "^0.0.0-alpha.31",
+        "@vitejs/devtools": "^0.1.0",
         "esbuild": "^0.27.0",
         "jiti": ">=1.21.0",
         "less": "^4.0.0",
@@ -1174,19 +1162,19 @@
       }
     },
     "node_modules/vitest": {
-      "version": "4.1.0",
-      "resolved": "https://registry.npmjs.org/vitest/-/vitest-4.1.0.tgz",
-      "integrity": "sha512-YbDrMF9jM2Lqc++2530UourxZHmkKLxrs4+mYhEwqWS97WJ7wOYEkcr+QfRgJ3PW9wz3odRijLZjHEaRLTNbqw==",
+      "version": "4.1.1",
+      "resolved": "https://registry.npmjs.org/vitest/-/vitest-4.1.1.tgz",
+      "integrity": "sha512-yF+o4POL41rpAzj5KVILUxm1GCjKnELvaqmU9TLLUbMfDzuN0UpUR9uaDs+mCtjPe+uYPksXDRLQGGPvj1cTmA==",
       "dev": true,
       "license": "MIT",
       "dependencies": {
-        "@vitest/expect": "4.1.0",
-        "@vitest/mocker": "4.1.0",
-        "@vitest/pretty-format": "4.1.0",
-        "@vitest/runner": "4.1.0",
-        "@vitest/snapshot": "4.1.0",
-        "@vitest/spy": "4.1.0",
-        "@vitest/utils": "4.1.0",
+        "@vitest/expect": "4.1.1",
+        "@vitest/mocker": "4.1.1",
+        "@vitest/pretty-format": "4.1.1",
+        "@vitest/runner": "4.1.1",
+        "@vitest/snapshot": "4.1.1",
+        "@vitest/spy": "4.1.1",
+        "@vitest/utils": "4.1.1",
         "es-module-lexer": "^2.0.0",
         "expect-type": "^1.3.0",
         "magic-string": "^0.30.21",
@@ -1198,7 +1186,7 @@
         "tinyexec": "^1.0.2",
         "tinyglobby": "^0.2.15",
         "tinyrainbow": "^3.0.3",
-        "vite": "^6.0.0 || ^7.0.0 || ^8.0.0-0",
+        "vite": "^6.0.0 || ^7.0.0 || ^8.0.0",
         "why-is-node-running": "^2.3.0"
       },
       "bin": {
@@ -1214,13 +1202,13 @@
         "@edge-runtime/vm": "*",
         "@opentelemetry/api": "^1.9.0",
         "@types/node": "^20.0.0 || ^22.0.0 || >=24.0.0",
-        "@vitest/browser-playwright": "4.1.0",
-        "@vitest/browser-preview": "4.1.0",
-        "@vitest/browser-webdriverio": "4.1.0",
-        "@vitest/ui": "4.1.0",
+        "@vitest/browser-playwright": "4.1.1",
+        "@vitest/browser-preview": "4.1.1",
+        "@vitest/browser-webdriverio": "4.1.1",
+        "@vitest/ui": "4.1.1",
         "happy-dom": "*",
         "jsdom": "*",
-        "vite": "^6.0.0 || ^7.0.0 || ^8.0.0-0"
+        "vite": "^6.0.0 || ^7.0.0 || ^8.0.0"
       },
       "peerDependenciesMeta": {
         "@edge-runtime/vm": {

sdk/typescript/package.json

@@ -32,7 +32,7 @@
   ],
   "license": "BUSL-1.1",
   "devDependencies": {
-    "typescript": "^5.4",
+    "typescript": "^6.0.0",
     "vitest": "^4.1"
   }
 }

sdk/typescript/src/base.ts

@@ -1,6 +1,15 @@
 import { TurnstoneAPIError } from "./errors.js";
 import { parseSSEStream } from "./sse.js";
 
+export interface TlsOptions {
+  /** Path to CA certificate PEM file (Node.js only). */
+  caCert?: string;
+  /** Path to client certificate PEM file for mTLS (Node.js only). */
+  clientCert?: string;
+  /** Path to client key PEM file for mTLS (Node.js only). */
+  clientKey?: string;
+}
+
 export interface ClientOptions {
   /** Server base URL (e.g. "http://localhost:8080"). */
   baseUrl: string;
@@ -8,6 +17,13 @@ export interface ClientOptions {
   token?: string;
   /** Custom fetch implementation (defaults to globalThis.fetch). */
   fetch?: typeof globalThis.fetch;
+  /**
+   * TLS certificate paths for documentation and tooling.
+   * The SDK does not read these directly — pass a custom `fetch`
+   * configured with your runtime's TLS agent (e.g. Node.js https.Agent).
+   * See docs/tls.md for examples.
+   */
+  tls?: TlsOptions;
 }
 
 export interface RequestOptions {

sdk/typescript/src/index.ts

@@ -19,7 +19,7 @@
 // Clients
 export { TurnstoneServer } from "./server.js";
 export { TurnstoneConsole } from "./console.js";
-export type { ClientOptions } from "./base.js";
+export type { ClientOptions, TlsOptions } from "./base.js";
 
 // Errors
 export { TurnstoneAPIError } from "./errors.js";

sdk/typescript/src/types.ts

@@ -186,7 +186,10 @@ export interface SkillInfo {
   agent_max_turns: number | null;
   notify_on_complete: string;
   enabled: boolean;
+  priority: number;
   allowed_tools: string;
+  license: string;
+  compatibility: string;
   resource_count: number;
   created: string;
   updated: string;
@@ -213,7 +216,10 @@ export interface CreateSkillRequest {
   agent_max_turns?: number | null;
   notify_on_complete?: string;
   enabled?: boolean;
+  priority?: number;
   allowed_tools?: string;
+  license?: string;
+  compatibility?: string;
 }
 
 export interface UpdateSkillRequest {
@@ -236,7 +242,10 @@ export interface UpdateSkillRequest {
   agent_max_turns?: number | null;
   notify_on_complete?: string;
   enabled?: boolean;
+  priority?: number;
   allowed_tools?: string;
+  license?: string;
+  compatibility?: string;
 }
 
 export interface ListSkillsResponse {
@@ -396,6 +405,7 @@ export interface ConsoleCreateWsRequest {
   model?: string;
   initial_message?: string;
   skill?: string;
+  resume_ws?: string;
 }
 
 export interface ConsoleCreateWsResponse {

tests.json

@@ -1,5 +1,5 @@
 {
-  "description": "turnstone behavior tests — tool selection, sequencing, and multi-step reasoning",
+  "description": "turnstone behavior tests \u2014 tool selection, sequencing, and multi-step reasoning",
   "defaults": {
     "n_runs": 5,
     "max_turns": 15
@@ -8,35 +8,91 @@
     {
       "id": "read-before-edit",
       "description": "Must read_file before edit_file on the same path",
-      "user_prompt": "Fix the typo in config.py — change 'recieve' to 'receive'",
+      "user_prompt": "Fix the typo in config.py \u2014 change 'recieve' to 'receive'",
       "setup": {
         "files": {
           "config.py": "# Config module\ndef recieve_data(source):\n    \"\"\"Recieve data from source.\"\"\"\n    return source.read()\n"
         }
       },
       "expected_actions": [
-        { "tool": "read_file", "args": { "path": "config.py" } },
-        { "tool": "edit_file", "args": { "path": "config.py" } }
+        {
+          "tool": "read_file",
+          "args": {
+            "path": "config.py"
+          }
+        },
+        {
+          "tool": "edit_file",
+          "args": {
+            "path": "config.py"
+          }
+        }
       ],
-      "match_mode": "ordered_subset"
+      "match_mode": "ordered_subset",
+      "user_prompts": [
+        "Fix the typo in config.py \u2014 change 'recieve' to 'receive'",
+        "In config.py, correct the misspelling of 'recieve' to 'receive'",
+        "Please update config.py by replacing 'recieve' with the correct spelling 'receive'",
+        "There's a typo in config.py: 'recieve' should be 'receive'. Please fix it.",
+        "Could you change 'recieve' to 'receive' in config.py?",
+        "Go ahead and fix 'recieve' \u2192 'receive' in config.py",
+        "I need the word 'recieve' corrected to 'receive' in the file config.py",
+        "config.py has a spelling error \u2014 'recieve' needs to be changed to 'receive'",
+        "Kindly rectify the typographical error in config.py, replacing 'recieve' with 'receive'",
+        "Hey, swap 'recieve' for 'receive' in config.py"
+      ]
     },
     {
       "id": "write-file-not-bash",
       "description": "Use write_file for file creation, not bash echo/cat",
       "user_prompt": "Create a file called hello.py that prints hello world",
       "expected_actions": [
-        { "tool": "write_file", "args_pattern": { "path": "hello\\.py" } }
+        {
+          "tool": "write_file",
+          "args_pattern": {
+            "path": "hello\\.py"
+          }
+        }
       ],
-      "match_mode": "subset"
+      "match_mode": "subset",
+      "user_prompts": [
+        "Create a file called hello.py that prints hello world",
+        "Make a hello.py file that outputs hello world",
+        "Please write a Python file named hello.py which prints hello world",
+        "I need a file called hello.py that prints hello world",
+        "Could you create hello.py with code that prints hello world?",
+        "Write hello.py \u2014 it should print hello world",
+        "Generate a hello.py file that outputs \"hello world\"",
+        "I'd like you to create a file named hello.py that prints hello world",
+        "Set up a file called hello.py to print hello world",
+        "Kindly produce a hello.py file whose purpose is to print hello world"
+      ]
     },
     {
       "id": "bash-for-commands",
       "description": "Use bash for running system commands",
       "user_prompt": "What Python version is installed?",
       "expected_actions": [
-        { "tool": "bash", "args_pattern": { "command": "python" } }
+        {
+          "tool": "bash",
+          "args_pattern": {
+            "command": "python"
+          }
+        }
       ],
-      "match_mode": "subset"
+      "match_mode": "subset",
+      "user_prompts": [
+        "What Python version is installed?",
+        "Check which version of Python is currently installed",
+        "Can you tell me the installed Python version?",
+        "python --version please",
+        "I need to know what version of Python is on this system",
+        "Which Python version do we have?",
+        "Could you look up the Python version that's installed here?",
+        "Determine the currently installed Python version",
+        "What's the Python version on this machine?",
+        "Please check the Python version"
+      ]
     },
     {
       "id": "search-for-patterns",
@@ -49,13 +105,30 @@
         }
       },
       "expected_actions": [
-        { "tool": "search", "args_pattern": { "query": "test_" } }
+        {
+          "tool": "search",
+          "args_pattern": {
+            "query": "test_"
+          }
+        }
       ],
-      "match_mode": "subset"
+      "match_mode": "subset",
+      "user_prompts": [
+        "Find all functions that start with 'test_' in the project",
+        "List every function in the project whose name begins with 'test_'",
+        "I need to locate all functions prefixed with 'test_' across the project",
+        "Could you search the project for any functions starting with 'test_'?",
+        "Show me all the test_ prefixed functions in this project",
+        "Hunt down every function that has a 'test_' prefix in the codebase",
+        "I'm looking for all functions named test_* throughout the project",
+        "Search the entire project for functions whose names start with test_",
+        "What functions beginning with 'test_' exist in this project?",
+        "Please identify all functions with the 'test_' prefix in the project files"
+      ]
     },
     {
       "id": "multi-file-edit",
-      "description": "Read and edit multiple files — must read before editing each, and edit both",
+      "description": "Read and edit multiple files \u2014 must read before editing each, and edit both",
       "user_prompt": "Change the default port from 8000 to 9000 in both server.py and config.py",
       "setup": {
         "files": {
@@ -64,12 +137,32 @@
         }
       },
       "expected_actions": [
-        { "tool": "read_file" },
-        { "tool": "read_file" },
-        { "tool": "edit_file" },
-        { "tool": "edit_file" }
+        {
+          "tool": "read_file"
+        },
+        {
+          "tool": "read_file"
+        },
+        {
+          "tool": "edit_file"
+        },
+        {
+          "tool": "edit_file"
+        }
       ],
-      "match_mode": "ordered_subset"
+      "match_mode": "ordered_subset",
+      "user_prompts": [
+        "Change the default port from 8000 to 9000 in both server.py and config.py",
+        "Update the default port to 9000 instead of 8000 in server.py and config.py",
+        "Could you modify the port number from 8000 to 9000 in both config.py and server.py?",
+        "Please replace port 8000 with 9000 in server.py and config.py",
+        "I need the default port switched from 8000 to 9000 in both server.py and config.py",
+        "In server.py and config.py, the default port should be changed from 8000 to 9000",
+        "Swap out port 8000 for 9000 in config.py and server.py",
+        "Would you mind updating the default port value from 8000 to 9000 across both server.py and config.py?",
+        "The default port in server.py and config.py needs to be 9000 instead of 8000 \u2014 please make that change",
+        "Go ahead and change 8000 to 9000 for the default port in both server.py and config.py"
+      ]
     },
     {
       "id": "search-then-edit",
@@ -83,51 +176,147 @@
         }
       },
       "expected_actions": [
-        { "tool": "search", "args_pattern": { "query": "MAX_RETRIES" } },
-        { "tool": "read_file" },
-        { "tool": "edit_file", "args_pattern": { "old_string": "3" } }
+        {
+          "tool": "search",
+          "args_pattern": {
+            "query": "MAX_RETRIES"
+          }
+        },
+        {
+          "tool": "read_file"
+        },
+        {
+          "tool": "edit_file",
+          "args_pattern": {
+            "old_string": "3"
+          }
+        }
       ],
-      "match_mode": "ordered_subset"
+      "match_mode": "ordered_subset",
+      "user_prompts": [
+        "Find where MAX_RETRIES is defined and change it from 3 to 5",
+        "Locate the definition of MAX_RETRIES and update its value from 3 to 5",
+        "Could you search for where MAX_RETRIES is defined and modify it from 3 to 5?",
+        "I need MAX_RETRIES changed from 3 to 5 \u2014 find where it's defined and update it",
+        "Please find the MAX_RETRIES definition and bump it from 3 to 5",
+        "Hunt down MAX_RETRIES in the codebase and change its value from 3 to 5",
+        "Where is MAX_RETRIES set to 3? Change it to 5.",
+        "Search the code for the MAX_RETRIES definition and alter it from 3 to 5",
+        "I'd like you to locate MAX_RETRIES (currently 3) and set it to 5 instead",
+        "Go find MAX_RETRIES and switch it from 3 to 5"
+      ]
     },
     {
       "id": "bash-git-log",
       "description": "Use bash for git commands, not other tools",
       "user_prompt": "Show me the git log for the last 5 commits",
       "expected_actions": [
-        { "tool": "bash", "args_pattern": { "command": "git\\s+log" } }
+        {
+          "tool": "bash",
+          "args_pattern": {
+            "command": "git\\s+log"
+          }
+        }
       ],
-      "match_mode": "subset"
+      "match_mode": "subset",
+      "user_prompts": [
+        "Show me the git log for the last 5 commits",
+        "Display the 5 most recent git commits",
+        "Can you pull up the git log limited to the last five commits?",
+        "I need to see the git log showing only the previous 5 commits",
+        "git log for the 5 latest commits, please",
+        "Would you mind showing me the last five entries in the git log?",
+        "Print out the most recent 5 commits from the git log",
+        "I'd like to review the git log \u2014 just the last 5 commits",
+        "Show the recent 5 commit history using git log",
+        "Could you display the git commit history for the past five commits?"
+      ]
     },
     {
       "id": "write-then-run",
       "description": "Create a script and run it to verify it works",
       "user_prompt": "Create a Python script called fib.py that prints the first 10 Fibonacci numbers, then run it to verify",
       "expected_actions": [
-        { "tool": "write_file", "args_pattern": { "path": "fib\\.py" } },
-        { "tool": "bash", "args_pattern": { "command": "python" } }
+        {
+          "tool": "write_file",
+          "args_pattern": {
+            "path": "fib\\.py"
+          }
+        },
+        {
+          "tool": "bash",
+          "args_pattern": {
+            "command": "python"
+          }
+        }
       ],
-      "match_mode": "ordered_subset"
+      "match_mode": "ordered_subset",
+      "user_prompts": [
+        "Create a Python script called fib.py that prints the first 10 Fibonacci numbers, then run it to verify",
+        "Write a Python file named fib.py that outputs the first 10 Fibonacci numbers, and execute it to confirm it works",
+        "Please make fib.py \u2013 a Python script printing the first ten Fibonacci numbers \u2013 then run it to check the output",
+        "I need a Python script fib.py that prints the first 10 Fibonacci numbers. Execute it afterwards to verify correctness.",
+        "Could you create fib.py to display the first 10 Fibonacci numbers in Python, and then run it to make sure it works?",
+        "Draft a script called fib.py in Python that outputs the first ten Fibonacci numbers, then execute it to validate",
+        "Hey, write me a fib.py that prints the first 10 Fibonacci numbers and run it so we can see it works",
+        "Generate a Python program fib.py which prints the initial 10 Fibonacci numbers, and verify by running it",
+        "Kindly produce a Python script named fib.py to print the first 10 Fibonacci numbers, then execute the script to confirm its output",
+        "Make a file fib.py containing Python code to print the first 10 Fibonacci numbers. Then run it to verify."
+      ]
     },
     {
       "id": "no-bash-for-file-write",
-      "description": "Should NOT use bash (echo/cat/heredoc) to create files — only write_file",
+      "description": "Should NOT use bash (echo/cat/heredoc) to create files \u2014 only write_file",
       "user_prompt": "Create a new file called README.md with a title and description of this project",
       "expected_actions": [
-        { "tool": "write_file", "args_pattern": { "path": "README" } }
+        {
+          "tool": "write_file",
+          "args_pattern": {
+            "path": "README"
+          }
+        }
       ],
-      "match_mode": "subset"
+      "match_mode": "subset",
+      "user_prompts": [
+        "Create a new file called README.md with a title and description of this project",
+        "Make a README.md file that includes a project title and description",
+        "I need a README.md created with a title and a brief description of the project",
+        "Please generate a README.md file containing the project's title and description",
+        "Could you set up a README.md with a title and project description?",
+        "Write a README.md that has a title and describes this project",
+        "Go ahead and create README.md \u2014 it should have a title and a description of the project",
+        "I'd like you to produce a new README.md file featuring a project title and description",
+        "Kindly establish a README.md file incorporating both a title and a description for this project",
+        "Spin up a README.md with a project title and description in it"
+      ]
     },
     {
-      "id": "plan-before-refactor",
-      "description": "Use the plan tool before a large refactoring task",
-      "user_prompt": "I need to refactor this codebase to separate the database layer from the API layer. Use the plan tool to think through the approach before making any changes.",
+      "id": "plan-when-asked",
+      "description": "Call the plan tool when the user asks to plan",
+      "user_prompt": "Plan how to add user authentication to this app.",
       "setup": {
         "files": {
-          "app.py": "import sqlite3\nfrom flask import Flask, jsonify\n\napp = Flask(__name__)\nDB = 'data.db'\n\ndef get_db():\n    return sqlite3.connect(DB)\n\n@app.route('/users')\ndef list_users():\n    db = get_db()\n    users = db.execute('SELECT * FROM users').fetchall()\n    db.close()\n    return jsonify(users)\n\n@app.route('/users/<int:uid>')\ndef get_user(uid):\n    db = get_db()\n    user = db.execute('SELECT * FROM users WHERE id=?', (uid,)).fetchone()\n    db.close()\n    return jsonify(user)\n\nif __name__ == '__main__':\n    app.run(port=8000)\n"
+          "app.py": "from flask import Flask, jsonify\n\napp = Flask(__name__)\n\n@app.route('/users')\ndef list_users():\n    return jsonify([{'id': 1, 'name': 'Alice'}])\n\nif __name__ == '__main__':\n    app.run(port=8000)\n"
         }
       },
-      "expected_actions": [{ "tool": "create_plan" }],
-      "match_mode": "subset"
+      "expected_actions": [
+        {
+          "tool": "plan_agent"
+        }
+      ],
+      "match_mode": "subset",
+      "user_prompts": [
+        "Plan how to add user authentication to this app.",
+        "Make a plan for adding pagination to the API endpoints.",
+        "Plan out how to add error handling to this application.",
+        "I need a plan for adding logging to this codebase.",
+        "Plan the approach for adding unit tests to this app.",
+        "How would you approach adding user authentication to this app? Lay out a plan.",
+        "I'd like you to outline a strategy for implementing user authentication in this application.",
+        "Could you come up with a plan for integrating user authentication into this app?",
+        "Think through the steps needed to add user auth to this app and present a plan.",
+        "Draft a plan for incorporating user authentication functionality into this application."
+      ]
     },
     {
       "id": "edit-not-rewrite",
@@ -139,10 +328,32 @@
         }
       },
       "expected_actions": [
-        { "tool": "read_file", "args": { "path": "utils.py" } },
-        { "tool": "edit_file", "args": { "path": "utils.py" } }
+        {
+          "tool": "read_file",
+          "args": {
+            "path": "utils.py"
+          }
+        },
+        {
+          "tool": "edit_file",
+          "args": {
+            "path": "utils.py"
+          }
+        }
       ],
-      "match_mode": "ordered_subset"
+      "match_mode": "ordered_subset",
+      "user_prompts": [
+        "Add a docstring to the process_data function in utils.py",
+        "Please add a docstring to the process_data function in utils.py",
+        "Could you write a docstring for process_data in utils.py?",
+        "Insert a docstring into the process_data function found in utils.py",
+        "I need a docstring added to process_data in utils.py",
+        "Put a docstring on the process_data function in utils.py",
+        "The process_data function in utils.py is missing a docstring \u2014 please add one",
+        "Would you mind adding a docstring to process_data in utils.py?",
+        "In utils.py, the process_data function needs a docstring",
+        "Add documentation via a docstring to the process_data function within utils.py"
+      ]
     },
     {
       "id": "bash-run-tests",
@@ -154,45 +365,130 @@
         }
       },
       "expected_actions": [
-        { "tool": "bash", "args_pattern": { "command": "pytest|python.*test" } }
+        {
+          "tool": "bash",
+          "args_pattern": {
+            "command": "pytest|python.*test"
+          }
+        }
       ],
-      "match_mode": "subset"
+      "match_mode": "subset",
+      "user_prompts": [
+        "Run the tests",
+        "Execute the test suite",
+        "Please go ahead and run the tests",
+        "Could you run the tests for me?",
+        "I need the tests to be run",
+        "Kick off the tests",
+        "Let's run the tests",
+        "Fire up the tests",
+        "Go ahead and execute the tests",
+        "I'd like you to run the tests"
+      ]
     },
     {
       "id": "web-fetch-url",
       "description": "Use web_fetch when asked to retrieve content from a URL",
       "user_prompt": "Fetch the contents of https://example.com and summarize what's on the page",
       "expected_actions": [
-        { "tool": "web_fetch", "args_pattern": { "url": "example\\.com" } }
+        {
+          "tool": "web_fetch",
+          "args_pattern": {
+            "url": "example\\.com"
+          }
+        }
       ],
-      "match_mode": "subset"
+      "match_mode": "subset",
+      "user_prompts": [
+        "Fetch the contents of https://example.com and summarize what's on the page",
+        "Go to https://example.com and give me a summary of what you find there",
+        "Could you pull up https://example.com and tell me what the page is about?",
+        "Retrieve the content from https://example.com, then provide a summary of it",
+        "I need you to grab https://example.com and summarize its contents for me",
+        "Please access https://example.com and give me an overview of the page",
+        "What's on https://example.com? Fetch it and summarize for me.",
+        "Download the page at https://example.com and provide a brief summary",
+        "I'd like a summary of whatever is at https://example.com \u2014 please fetch it first",
+        "Hit https://example.com and let me know what's there in summary form"
+      ]
     },
     {
       "id": "man-page-lookup",
       "description": "Use man tool to look up command documentation",
       "user_prompt": "Look up the man page for tar and tell me what the --xattrs flag does",
       "expected_actions": [
-        { "tool": "man", "args_pattern": { "page": "tar" } }
+        {
+          "tool": "man",
+          "args_pattern": {
+            "page": "tar"
+          }
+        }
       ],
-      "match_mode": "subset"
+      "match_mode": "subset",
+      "user_prompts": [
+        "Look up the man page for tar and tell me what the --xattrs flag does",
+        "What does the --xattrs flag do in tar? Check the man page for me.",
+        "Could you pull up the man page for tar and explain the --xattrs option?",
+        "I need to know what --xattrs does in tar \u2014 can you check the man page?",
+        "Check tar's man page and let me know the purpose of the --xattrs flag.",
+        "Please consult the tar man page and describe what the --xattrs flag is for.",
+        "Hey, look at the tar man page real quick \u2014 what's --xattrs do?",
+        "I'd like you to read the tar man page and summarize the --xattrs option for me.",
+        "Would you mind checking the man page for tar to find out what --xattrs means?",
+        "Look into the tar manual and explain the --xattrs flag to me."
+      ]
     },
     {
       "id": "math-calculation",
       "description": "Use the math tool for precise calculations, not bash or mental math",
       "user_prompt": "What is 2^64 - 1? Use the math tool to calculate it precisely.",
       "expected_actions": [
-        { "tool": "math", "args_pattern": { "code": "2.*64" } }
+        {
+          "tool": "math",
+          "args_pattern": {
+            "code": "2.*64"
+          }
+        }
       ],
-      "match_mode": "subset"
+      "match_mode": "subset",
+      "user_prompts": [
+        "What is 2^64 - 1? Use the math tool to calculate it precisely.",
+        "Calculate 2^64 - 1 for me using the math tool, please.",
+        "I need the exact value of 2^64 - 1. Please use the math tool.",
+        "Could you use the math tool to compute 2^64 minus 1 precisely?",
+        "Use the math tool to tell me what 2^64 - 1 equals.",
+        "I'm curious: what's 2^64 - 1? Compute it with the math tool.",
+        "Please precisely determine 2^64 - 1 via the math tool.",
+        "Mind using the math tool to figure out 2^64 - 1 exactly?",
+        "I'd like to know the precise result of 2^64 - 1 \u2014 use the math tool for this.",
+        "Leverage the math tool to give me an exact answer for 2^64 - 1."
+      ]
     },
     {
       "id": "web-search-query",
       "description": "Use web_search for general knowledge lookups, not web_fetch",
       "user_prompt": "Search the web for the current population of Tokyo",
       "expected_actions": [
-        { "tool": "web_search", "args_pattern": { "query": "Tokyo" } }
+        {
+          "tool": "web_search",
+          "args_pattern": {
+            "query": "Tokyo"
+          }
+        }
       ],
-      "match_mode": "subset"
+      "match_mode": "subset",
+      "user_prompts": [
+        "Search the web for the current population of Tokyo",
+        "What's Tokyo's current population? Look it up on the web.",
+        "Could you do a web search to find out how many people currently live in Tokyo?",
+        "Please search online for Tokyo's present-day population.",
+        "I need you to look up the current population of Tokyo on the web.",
+        "Find me Tokyo's current population via a web search.",
+        "Web search: what is the current population of Tokyo?",
+        "I'd like to know Tokyo's current population\u2014can you search the web for that?",
+        "Look up how many people live in Tokyo right now using a web search.",
+        "Do a web search for the population of Tokyo as of now."
+      ]
     }
   ]
 }

tests/conftest.py

@@ -1,11 +1,23 @@
+from __future__ import annotations
+
+import os
 from unittest.mock import MagicMock
 
 import pytest
 
 
+def pytest_addoption(parser: pytest.Parser) -> None:
+    parser.addoption(
+        "--storage-backend",
+        default="sqlite",
+        choices=["sqlite", "postgresql"],
+        help="Storage backend for integration tests (default: sqlite)",
+    )
+
+
 @pytest.fixture
 def tmp_db(tmp_path):
-    """Provide a temporary SQLite storage backend."""
+    """Provide a temporary SQLite storage backend (singleton registry)."""
     from turnstone.core.storage import init_storage, reset_storage
 
     db_path = str(tmp_path / "test.db")
@@ -15,6 +27,68 @@ def tmp_db(tmp_path):
     reset_storage()
 
 
+@pytest.fixture
+def storage_backend(request, tmp_path):
+    """Shared storage backend fixture — respects --storage-backend flag.
+
+    Returns a StorageBackend instance (SQLite or PostgreSQL).
+    Tests that use this fixture run against whichever backend CI selects.
+    """
+    from turnstone.core.storage import init_storage, reset_storage
+
+    backend_type = request.config.getoption("--storage-backend")
+    reset_storage()
+
+    if backend_type == "postgresql":
+        pg_url = os.environ.get(
+            "TURNSTONE_TEST_PG_URL",
+            "postgresql+psycopg://postgres:postgres@localhost:5432/turnstone_test",
+        )
+        backend = init_storage("postgresql", url=pg_url, run_migrations=False)
+        yield backend
+        # Truncate all tables between tests — faster than DELETE and resets
+        # autoincrement sequences.  CASCADE handles any future FK constraints.
+        # NOTE: accesses backend._engine (SQLAlchemy internal) — both SQLite
+        # and PostgreSQL backends expose this.  If a non-SQLAlchemy backend is
+        # ever added, this cleanup will need a protocol-level hook.
+        try:
+            import sqlalchemy as sa
+
+            from turnstone.core.storage._schema import metadata as db_metadata
+
+            with backend._engine.connect() as conn:
+                table_names = ", ".join(t.name for t in reversed(db_metadata.sorted_tables))
+                conn.execute(sa.text(f"TRUNCATE {table_names} RESTART IDENTITY CASCADE"))
+                conn.commit()
+        except Exception:
+            pass  # best-effort cleanup; reset_storage disposes engine
+        finally:
+            reset_storage()
+    else:
+        db_path = str(tmp_path / "test.db")
+        backend = init_storage("sqlite", path=db_path, run_migrations=False)
+        yield backend
+        reset_storage()
+
+
+@pytest.fixture
+def backend(storage_backend):
+    """Alias for storage_backend — used by test_storage_sqlite.py etc."""
+    return storage_backend
+
+
+@pytest.fixture
+def db(storage_backend):
+    """Alias for storage_backend — used by domain-specific storage tests."""
+    return storage_backend
+
+
+@pytest.fixture
+def storage(storage_backend):
+    """Alias for storage_backend — used by services/skill resource tests."""
+    return storage_backend
+
+
 @pytest.fixture
 def mock_openai_client():
     """Return a minimal mock OpenAI client."""

tests/test_api_versioning.py

@@ -19,6 +19,7 @@ class TestServerVersioning:
 
         mock_mgr = MagicMock()
         mock_mgr.list_all.return_value = []
+        mock_mgr.max_workstreams = 10
         app = create_app(
             workstreams=mock_mgr,
             global_queue=queue.Queue(),

tests/test_auth.py

@@ -783,6 +783,7 @@ class TestServerAuth:
         mock_ws.session = mock_session
         mock_mgr = MagicMock()
         mock_mgr.list_all.return_value = [mock_ws]
+        mock_mgr.max_workstreams = 10
 
         app = srv_mod.create_app(
             workstreams=mock_mgr,
@@ -1001,6 +1002,7 @@ class TestServerLogin:
         mock_ws.session = mock_session
         mock_mgr = MagicMock()
         mock_mgr.list_all.return_value = [mock_ws]
+        mock_mgr.max_workstreams = 10
 
         app = srv_mod.create_app(
             workstreams=mock_mgr,
@@ -1235,6 +1237,52 @@ class TestJWTAudienceIssuer:
         result = validate_jwt(token, self.SECRET, audience="")
         assert result is not None
 
+    def test_create_jwt_expiry_seconds(self):
+        import jwt as pyjwt
+
+        from turnstone.core.auth import create_jwt
+
+        token = create_jwt("user1", frozenset({"read"}), "test", self.SECRET, expiry_seconds=300)
+        payload = pyjwt.decode(
+            token, self.SECRET, algorithms=["HS256"], options={"verify_aud": False}
+        )
+        assert payload["exp"] - payload["iat"] == 300
+
+    def test_create_jwt_expiry_seconds_overrides_hours(self):
+        import jwt as pyjwt
+
+        from turnstone.core.auth import create_jwt
+
+        token = create_jwt(
+            "user1",
+            frozenset({"read"}),
+            "test",
+            self.SECRET,
+            expiry_hours=24,
+            expiry_seconds=60,
+        )
+        payload = pyjwt.decode(
+            token, self.SECRET, algorithms=["HS256"], options={"verify_aud": False}
+        )
+        # expiry_seconds takes precedence over expiry_hours
+        assert payload["exp"] - payload["iat"] == 60
+
+    def test_create_jwt_expiry_seconds_rejects_zero(self):
+        import pytest
+
+        from turnstone.core.auth import create_jwt
+
+        with pytest.raises(ValueError, match="expiry_seconds must be positive"):
+            create_jwt("user1", frozenset({"read"}), "test", self.SECRET, expiry_seconds=0)
+
+    def test_create_jwt_expiry_seconds_rejects_negative(self):
+        import pytest
+
+        from turnstone.core.auth import create_jwt
+
+        with pytest.raises(ValueError, match="expiry_seconds must be positive"):
+            create_jwt("user1", frozenset({"read"}), "test", self.SECRET, expiry_seconds=-1)
+
 
 class TestServiceTokenManager:
     SECRET = "test-secret-that-is-at-least-32-chars"
@@ -1369,8 +1417,11 @@ class TestCorsConfigurable:
 
         import turnstone.server as srv_mod
 
+        mgr = MagicMock()
+        mgr.list_all.return_value = []
+        mgr.max_workstreams = 10
         app = srv_mod.create_app(
-            workstreams=MagicMock(),
+            workstreams=mgr,
             global_queue=queue.Queue(),
             global_listeners=[],
             global_listeners_lock=threading.Lock(),
@@ -1388,8 +1439,11 @@ class TestCorsConfigurable:
 
         import turnstone.server as srv_mod
 
+        mgr = MagicMock()
+        mgr.list_all.return_value = []
+        mgr.max_workstreams = 10
         app = srv_mod.create_app(
-            workstreams=MagicMock(),
+            workstreams=mgr,
             global_queue=queue.Queue(),
             global_listeners=[],
             global_listeners_lock=threading.Lock(),

tests/test_auth_identity.py

@@ -130,7 +130,7 @@ class TestParseScopes:
 
 
 class TestJWT:
-    SECRET = "test-secret-key-for-jwt"
+    SECRET = "test-secret-key-for-jwt-min-32b!"
 
     def test_round_trip(self):
         scopes = frozenset({"read", "write"})
@@ -155,7 +155,7 @@ class TestJWT:
 
     def test_invalid_signature(self):
         token = create_jwt("user1", frozenset({"read"}), "db", self.SECRET)
-        assert validate_jwt(token, "wrong-secret") is None
+        assert validate_jwt(token, "wrong-secret-key-for-jwt-min-32b") is None
 
     def test_malformed_token(self):
         assert validate_jwt("not.a.jwt", self.SECRET) is None
@@ -217,7 +217,7 @@ class TestAuthenticateToken:
         assert result.scopes == frozenset({"read", "write", "approve"})
 
     def test_jwt_token(self):
-        secret = "test-secret"
+        secret = "test-secret-key-for-jwt-min-32b!"
         jwt_tok = create_jwt("user1", frozenset({"read", "write"}), "db", secret)
         cfg = AuthConfig(enabled=True)
         result = _authenticate_token(jwt_tok, cfg, jwt_secret=secret)
@@ -304,7 +304,7 @@ class TestCheckRequestScopes:
         assert result.has_scope("approve")
 
     def test_jwt_with_scopes(self):
-        secret = "test"
+        secret = "test-secret-key-for-jwt-min-32b!"
         jwt_tok = create_jwt("u1", frozenset({"read", "write"}), "db", secret)
         cfg = AuthConfig(enabled=True)
         allowed, status, msg, result = check_request(
@@ -319,7 +319,7 @@ class TestCheckRequestScopes:
         assert result.user_id == "u1"
 
     def test_jwt_insufficient_scope(self):
-        secret = "test"
+        secret = "test-secret-key-for-jwt-min-32b!"
         jwt_tok = create_jwt("u1", frozenset({"read"}), "db", secret)
         cfg = AuthConfig(enabled=True)
         allowed, status, msg, _ = check_request(

tests/test_bridge_stress.py

@@ -0,0 +1,351 @@
+"""Stress tests for bridge.py threading — race conditions in approval,
+plan review, and workstream lifecycle.
+
+Each scenario is run many times (ITERATIONS) with threading.Barrier to
+maximize timing overlap.  Uses mock broker (no Redis) and no HTTP calls.
+
+Races tested:
+1. Duplicate approval on SSE reconnect (TOCTOU in _pending_approvals)
+2. Duplicate plan review on SSE reconnect (TOCTOU in _pending_plan_reviews)
+3. approve_set stale reference escape during concurrent update
+4. _running flag visibility across threads on shutdown
+5. Approval thread exits within bounded time after timeout
+6. Concurrent approval + workstream close leaves no orphaned state
+"""
+
+from __future__ import annotations
+
+import threading
+import time
+from collections import Counter
+from unittest.mock import MagicMock, patch
+
+from turnstone.mq.bridge import Bridge
+
+ITERATIONS = 100
+
+# ---------------------------------------------------------------------------
+# Helpers
+# ---------------------------------------------------------------------------
+
+
+def _make_bridge(**overrides) -> Bridge:
+    """Create a Bridge with a mock broker (no Redis or HTTP)."""
+    broker = MagicMock()
+    defaults = dict(
+        server_url="http://localhost:8080",
+        broker=broker,
+        node_id="test-node",
+        approval_timeout=1,
+    )
+    defaults.update(overrides)
+    return Bridge(**defaults)
+
+
+def _approval_items(tool_name: str = "bash") -> list[dict]:
+    return [{"func_name": tool_name, "needs_approval": True, "approval_label": tool_name}]
+
+
+def _wait_pending_resolved(bridge: Bridge, key: str, attr: str, deadline_s: float = 3.0) -> bool:
+    """Poll until the pending entry is resolved (tombstone) or absent."""
+    deadline = time.monotonic() + deadline_s
+    while time.monotonic() < deadline:
+        with bridge._lock:
+            entries = getattr(bridge, attr)
+            if key not in entries:
+                return True
+            _, resolved_at = entries[key]
+            if resolved_at > 0:
+                return True
+        time.sleep(0.01)
+    return False
+
+
+# ---------------------------------------------------------------------------
+# Race 1: Duplicate approval on SSE reconnect
+# ---------------------------------------------------------------------------
+
+
+class TestDuplicateApproval:
+    """Two threads call _handle_approval for the same ws_id simultaneously.
+    Only one should create a pending entry; the other should be skipped."""
+
+    def test_no_duplicate_approvals(self):
+        sent_count = Counter()
+
+        for _ in range(ITERATIONS):
+            bridge = _make_bridge()
+            bridge._broker.pop_response.return_value = '{"type": "approve", "approved": true}'
+            barrier = threading.Barrier(2, timeout=5)
+
+            def _call_approval(bridge=bridge, barrier=barrier):
+                barrier.wait()
+                bridge._handle_approval("ws-1", {"items": _approval_items()})
+
+            t1 = threading.Thread(target=_call_approval)
+            t2 = threading.Thread(target=_call_approval)
+            with (
+                patch.object(bridge, "_api_approve") as mock_approve,
+                patch.object(bridge, "_publish_ws"),
+            ):
+                t1.start()
+                t2.start()
+                t1.join(timeout=5)
+                t2.join(timeout=5)
+                assert not t1.is_alive(), "Thread 1 hung"
+                assert not t2.is_alive(), "Thread 2 hung"
+
+                # Wait for spawned _wait_approval threads to resolve
+                _wait_pending_resolved(bridge, "ws-1", "_pending_approvals")
+
+                sent_count[mock_approve.call_count] += 1
+
+        # At most 1 approval should be forwarded per iteration
+        assert sent_count.get(2, 0) == 0, (
+            f"Duplicate approvals sent in {sent_count[2]}/{ITERATIONS} iterations"
+        )
+
+
+# ---------------------------------------------------------------------------
+# Race 2: Duplicate plan review on SSE reconnect
+# ---------------------------------------------------------------------------
+
+
+class TestDuplicatePlanReview:
+    """Two threads call _handle_plan_review simultaneously.
+    Only one should create a pending entry."""
+
+    def test_no_duplicate_plan_reviews(self):
+        sent_count = Counter()
+
+        for _ in range(ITERATIONS):
+            bridge = _make_bridge()
+            bridge._broker.pop_response.return_value = (
+                '{"type": "plan_feedback", "feedback": "looks good"}'
+            )
+            barrier = threading.Barrier(2, timeout=5)
+
+            def _call_plan(bridge=bridge, barrier=barrier):
+                barrier.wait()
+                bridge._handle_plan_review("ws-1", {"content": "plan text"})
+
+            t1 = threading.Thread(target=_call_plan)
+            t2 = threading.Thread(target=_call_plan)
+            with patch.object(bridge, "_publish_ws"), patch.object(bridge._http, "post"):
+                t1.start()
+                t2.start()
+                t1.join(timeout=5)
+                t2.join(timeout=5)
+                assert not t1.is_alive(), "Thread 1 hung"
+                assert not t2.is_alive(), "Thread 2 hung"
+
+                # Wait for spawned _wait_plan threads to resolve
+                _wait_pending_resolved(bridge, "ws-1", "_pending_plan_reviews")
+
+                sent_count[bridge._http.post.call_count] += 1
+
+        assert sent_count.get(2, 0) == 0, (
+            f"Duplicate plan reviews sent in {sent_count[2]}/{ITERATIONS} iterations"
+        )
+
+
+# ---------------------------------------------------------------------------
+# Race 3: approve_set stale reference during concurrent update
+# ---------------------------------------------------------------------------
+
+
+class TestApproveSetConsistency:
+    """One thread reads approve_set for auto-approve check while another
+    updates it via _wait_approval 'always' path.  The auto-approve
+    decision should be consistent (either all-approved or not)."""
+
+    def test_approve_set_never_partially_visible(self):
+        for _ in range(ITERATIONS):
+            bridge = _make_bridge()
+            with bridge._lock:
+                bridge._ws_approve_tools["ws-1"] = {"read_file", "search"}
+
+            barrier = threading.Barrier(2, timeout=5)
+            results = []
+
+            def _reader(bridge=bridge, barrier=barrier, results=results):
+                barrier.wait()
+                with bridge._lock:
+                    snap = bridge._ws_approve_tools.get("ws-1", set()).copy()
+                results.append(snap)
+
+            def _writer(bridge=bridge, barrier=barrier):
+                barrier.wait()
+                with bridge._lock:
+                    existing = bridge._ws_approve_tools.get("ws-1", set())
+                    bridge._ws_approve_tools["ws-1"] = existing | {"bash", "write_file"}
+
+            t1 = threading.Thread(target=_reader)
+            t2 = threading.Thread(target=_writer)
+            t1.start()
+            t2.start()
+            t1.join(timeout=5)
+            t2.join(timeout=5)
+            assert not t1.is_alive(), "Reader hung"
+            assert not t2.is_alive(), "Writer hung"
+
+            snap = results[0]
+            assert snap in (
+                {"read_file", "search"},
+                {"read_file", "search", "bash", "write_file"},
+            ), f"Partial set observed: {snap}"
+
+
+# ---------------------------------------------------------------------------
+# Race 4: _running flag visibility across threads
+# ---------------------------------------------------------------------------
+
+
+class TestRunningFlagVisibility:
+    """All threads reading _running should see False within a bounded time
+    after the main thread sets it."""
+
+    def test_all_threads_observe_shutdown(self):
+        bridge = _make_bridge()
+        observed_false = threading.Event()
+        threads_running = []
+
+        def _spin_checker():
+            while bridge._running:
+                time.sleep(0.001)
+            observed_false.set()
+
+        for _ in range(5):
+            t = threading.Thread(target=_spin_checker, daemon=True)
+            threads_running.append(t)
+            t.start()
+
+        time.sleep(0.01)
+        bridge._running = False
+
+        for t in threads_running:
+            t.join(timeout=1)
+            assert not t.is_alive(), "Thread did not observe _running=False"
+
+        assert observed_false.is_set()
+
+
+# ---------------------------------------------------------------------------
+# Race 5: Approval thread exits within bounded time
+# ---------------------------------------------------------------------------
+
+
+class TestApprovalThreadTimeout:
+    """An approval thread blocked on pop_response should exit within the
+    configured approval_timeout, not hang indefinitely."""
+
+    def test_approval_thread_exits_within_timeout(self):
+        for _ in range(10):
+            bridge = _make_bridge(approval_timeout=0.5)
+
+            def _slow_pop(queue_name, timeout=300):
+                time.sleep(min(timeout, 0.5))
+                return None
+
+            bridge._broker.pop_response.side_effect = _slow_pop
+
+            with patch.object(bridge, "_publish_ws"), patch.object(bridge, "_api_approve"):
+                bridge._handle_approval("ws-1", {"items": _approval_items()})
+
+            # The pending entry should be resolved within the timeout
+            resolved = _wait_pending_resolved(bridge, "ws-1", "_pending_approvals", deadline_s=3.0)
+            assert resolved, "Approval thread did not exit within expected timeout"
+
+
+# ---------------------------------------------------------------------------
+# Race 6: Concurrent approval + workstream close
+# ---------------------------------------------------------------------------
+
+
+class TestApprovalDuringClose:
+    """An approval arriving at the exact same time as a ws_closed event
+    should not leave orphaned state."""
+
+    def test_no_orphaned_pending_after_close(self):
+        for _ in range(ITERATIONS):
+            bridge = _make_bridge(approval_timeout=0.1)
+            bridge._broker.pop_response.return_value = None  # timeout
+
+            barrier = threading.Barrier(2, timeout=5)
+
+            def _send_approval(bridge=bridge, barrier=barrier):
+                barrier.wait()
+                with patch.object(bridge, "_publish_ws"), patch.object(bridge, "_api_approve"):
+                    bridge._handle_approval("ws-1", {"items": _approval_items()})
+
+            def _close_ws(bridge=bridge, barrier=barrier):
+                barrier.wait()
+                with (
+                    patch.object(bridge, "_publish_global"),
+                    patch.object(bridge, "_publish_cluster"),
+                ):
+                    bridge._handle_global_event({"type": "ws_closed", "ws_id": "ws-1"})
+
+            t1 = threading.Thread(target=_send_approval)
+            t2 = threading.Thread(target=_close_ws)
+            t1.start()
+            t2.start()
+            t1.join(timeout=5)
+            t2.join(timeout=5)
+            assert not t1.is_alive(), "Approval thread hung"
+            assert not t2.is_alive(), "Close thread hung"
+
+            # Wait for spawned _wait_approval thread to resolve (if close
+            # didn't remove the entry first)
+            resolved = _wait_pending_resolved(bridge, "ws-1", "_pending_approvals")
+            assert resolved, "Orphaned pending approval"
+
+
+# ---------------------------------------------------------------------------
+# Race 7: Plan review refinement loop (tombstone → cleanup → re-entry)
+# ---------------------------------------------------------------------------
+
+
+class TestPlanReviewRefinementLoop:
+    """After a plan review is resolved, a ws_state event should clean up the
+    tombstone so the refinement-loop plan_review event is handled correctly."""
+
+    def test_refinement_loop_allows_reentry(self):
+        for _ in range(ITERATIONS):
+            bridge = _make_bridge()
+            bridge._broker.pop_response.return_value = (
+                '{"type": "plan_feedback", "feedback": "refine this"}'
+            )
+
+            # Step 1: first plan review — creates pending entry, resolves it
+            with patch.object(bridge, "_publish_ws"), patch.object(bridge._http, "post"):
+                bridge._handle_plan_review("ws-1", {"content": "plan v1"})
+
+            _wait_pending_resolved(bridge, "ws-1", "_pending_plan_reviews")
+
+            # Verify tombstone is present (resolved_at > 0)
+            with bridge._lock:
+                assert "ws-1" in bridge._pending_plan_reviews
+                assert bridge._pending_plan_reviews["ws-1"][1] > 0
+
+            # Step 2: ws_state event cleans up the resolved tombstone
+            with (
+                patch.object(bridge, "_publish_ws"),
+                patch.object(bridge, "_publish_global"),
+                patch.object(bridge, "_publish_cluster"),
+            ):
+                bridge._handle_global_event(
+                    {"type": "ws_state", "ws_id": "ws-1", "state": "working"}
+                )
+
+            with bridge._lock:
+                assert "ws-1" not in bridge._pending_plan_reviews
+
+            # Step 3: refinement plan_review arrives — should create new entry
+            with patch.object(bridge, "_publish_ws"), patch.object(bridge._http, "post"):
+                bridge._handle_plan_review("ws-1", {"content": "plan v2"})
+
+            _wait_pending_resolved(bridge, "ws-1", "_pending_plan_reviews")
+
+            with bridge._lock:
+                assert "ws-1" in bridge._pending_plan_reviews

tests/test_channel_storage.py

@@ -2,17 +2,6 @@
 
 from __future__ import annotations
 
-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
-
 
 class TestChannelUserCRUD:
     """Tests for channel_users table operations."""

tests/test_config.py

@@ -3,54 +3,55 @@
 import argparse
 
 import turnstone.core.config as config_mod
-from turnstone.core.config import apply_config, load_config
+from turnstone.core.config import apply_config, load_config, set_config_path
 
 
 def _reset_cache():
     """Clear the module-level config cache between tests."""
     config_mod._cache = None
+    config_mod._config_path = None
 
 
-def test_load_config_missing_file(tmp_path, monkeypatch):
+def test_load_config_missing_file(tmp_path):
     _reset_cache()
-    monkeypatch.setattr(config_mod, "CONFIG_PATH", tmp_path / "nope.toml")
+    set_config_path(str(tmp_path / "nope.toml"))
     assert load_config() == {}
 
 
-def test_load_config_valid_toml(tmp_path, monkeypatch):
+def test_load_config_valid_toml(tmp_path):
     _reset_cache()
     cfg = tmp_path / "config.toml"
     cfg.write_text('[redis]\nhost = "10.0.0.1"\nport = 6380\npassword = "secret"\n')
-    monkeypatch.setattr(config_mod, "CONFIG_PATH", cfg)
+    set_config_path(str(cfg))
     result = load_config()
     assert result["redis"]["host"] == "10.0.0.1"
     assert result["redis"]["port"] == 6380
     assert result["redis"]["password"] == "secret"
 
 
-def test_load_config_section(tmp_path, monkeypatch):
+def test_load_config_section(tmp_path):
     _reset_cache()
     cfg = tmp_path / "config.toml"
     cfg.write_text('[api]\nbase_url = "http://x:8000/v1"\n[redis]\nhost = "y"\n')
-    monkeypatch.setattr(config_mod, "CONFIG_PATH", cfg)
+    set_config_path(str(cfg))
     assert load_config("redis") == {"host": "y"}
     assert load_config("api") == {"base_url": "http://x:8000/v1"}
     assert load_config("nonexistent") == {}
 
 
-def test_load_config_invalid_toml(tmp_path, monkeypatch):
+def test_load_config_invalid_toml(tmp_path):
     _reset_cache()
     cfg = tmp_path / "config.toml"
     cfg.write_text("this is not valid toml [[[")
-    monkeypatch.setattr(config_mod, "CONFIG_PATH", cfg)
+    set_config_path(str(cfg))
     assert load_config() == {}
 
 
-def test_load_config_caches(tmp_path, monkeypatch):
+def test_load_config_caches(tmp_path):
     _reset_cache()
     cfg = tmp_path / "config.toml"
     cfg.write_text('[api]\nbase_url = "http://first"\n')
-    monkeypatch.setattr(config_mod, "CONFIG_PATH", cfg)
+    set_config_path(str(cfg))
     first = load_config()
     assert first["api"]["base_url"] == "http://first"
 
@@ -60,14 +61,14 @@ def test_load_config_caches(tmp_path, monkeypatch):
     assert second["api"]["base_url"] == "http://first"
 
 
-def test_apply_config_sets_defaults(tmp_path, monkeypatch):
+def test_apply_config_sets_defaults(tmp_path):
     _reset_cache()
     cfg = tmp_path / "config.toml"
     cfg.write_text(
         '[redis]\nhost = "redis.local"\nport = 7777\npassword = "pw"\n'
         '[bridge]\nserver_url = "http://bridge:9090"\n'
     )
-    monkeypatch.setattr(config_mod, "CONFIG_PATH", cfg)
+    set_config_path(str(cfg))
 
     parser = argparse.ArgumentParser()
     parser.add_argument("--redis-host", default="localhost")
@@ -84,11 +85,11 @@ def test_apply_config_sets_defaults(tmp_path, monkeypatch):
     assert args.server_url == "http://bridge:9090"
 
 
-def test_apply_config_cli_overrides(tmp_path, monkeypatch):
+def test_apply_config_cli_overrides(tmp_path):
     _reset_cache()
     cfg = tmp_path / "config.toml"
     cfg.write_text('[redis]\nhost = "config-host"\nport = 7777\n')
-    monkeypatch.setattr(config_mod, "CONFIG_PATH", cfg)
+    set_config_path(str(cfg))
 
     parser = argparse.ArgumentParser()
     parser.add_argument("--redis-host", default="localhost")
@@ -102,11 +103,11 @@ def test_apply_config_cli_overrides(tmp_path, monkeypatch):
     assert args.redis_port == 7777  # config wins (no CLI override)
 
 
-def test_apply_config_missing_keys_keep_defaults(tmp_path, monkeypatch):
+def test_apply_config_missing_keys_keep_defaults(tmp_path):
     _reset_cache()
     cfg = tmp_path / "config.toml"
     cfg.write_text('[redis]\nhost = "only-host"\n')  # no port, no password
-    monkeypatch.setattr(config_mod, "CONFIG_PATH", cfg)
+    set_config_path(str(cfg))
 
     parser = argparse.ArgumentParser()
     parser.add_argument("--redis-host", default="localhost")
@@ -121,9 +122,9 @@ def test_apply_config_missing_keys_keep_defaults(tmp_path, monkeypatch):
     assert args.redis_password is None  # original default kept
 
 
-def test_apply_config_no_file(tmp_path, monkeypatch):
+def test_apply_config_no_file(tmp_path):
     _reset_cache()
-    monkeypatch.setattr(config_mod, "CONFIG_PATH", tmp_path / "nope.toml")
+    set_config_path(str(tmp_path / "nope.toml"))
 
     parser = argparse.ArgumentParser()
     parser.add_argument("--redis-host", default="localhost")
@@ -133,11 +134,11 @@ def test_apply_config_no_file(tmp_path, monkeypatch):
     assert args.redis_host == "localhost"
 
 
-def test_apply_config_model_section(tmp_path, monkeypatch):
+def test_apply_config_model_section(tmp_path):
     _reset_cache()
     cfg = tmp_path / "config.toml"
     cfg.write_text('[model]\nname = "qwen-72b"\ntemperature = 0.3\n')
-    monkeypatch.setattr(config_mod, "CONFIG_PATH", cfg)
+    set_config_path(str(cfg))
 
     parser = argparse.ArgumentParser()
     parser.add_argument("--model", default=None)
@@ -158,7 +159,7 @@ def test_tavily_key_from_config(tmp_path, monkeypatch):
 
     cfg = tmp_path / "config.toml"
     cfg.write_text('[api]\ntavily_key = "tvly-from-config"\n')
-    monkeypatch.setattr(config_mod, "CONFIG_PATH", cfg)
+    set_config_path(str(cfg))
     monkeypatch.delenv("TAVILY_API_KEY", raising=False)
 
     key = config_mod.get_tavily_key()
@@ -174,14 +175,14 @@ def test_tavily_key_fallback_to_env(tmp_path, monkeypatch):
     # Config exists but no tavily_key in it
     cfg = tmp_path / "config.toml"
     cfg.write_text("[api]\n")
-    monkeypatch.setattr(config_mod, "CONFIG_PATH", cfg)
+    set_config_path(str(cfg))
     monkeypatch.setenv("TAVILY_API_KEY", "tvly-from-env")
 
     key = config_mod.get_tavily_key()
     assert key == "tvly-from-env"
 
 
-def test_apply_config_judge_section(tmp_path, monkeypatch):
+def test_apply_config_judge_section(tmp_path):
     """apply_config() loads [judge] section and maps to argparse dests."""
     _reset_cache()
     cfg = tmp_path / "config.toml"
@@ -193,7 +194,7 @@ def test_apply_config_judge_section(tmp_path, monkeypatch):
         "timeout = 30.0\n"
         "read_only_tools = false\n"
     )
-    monkeypatch.setattr(config_mod, "CONFIG_PATH", cfg)
+    set_config_path(str(cfg))
 
     parser = argparse.ArgumentParser()
     parser.add_argument("--judge", dest="judge_enabled", action="store_true", default=False)
@@ -212,12 +213,12 @@ def test_apply_config_judge_section(tmp_path, monkeypatch):
     assert args.judge_read_only_tools is False
 
 
-def test_apply_config_judge_cli_overrides(tmp_path, monkeypatch):
+def test_apply_config_judge_cli_overrides(tmp_path):
     """CLI flags override config.toml [judge] values."""
     _reset_cache()
     cfg = tmp_path / "config.toml"
     cfg.write_text("[judge]\nenabled = true\nconfidence_threshold = 0.85\n")
-    monkeypatch.setattr(config_mod, "CONFIG_PATH", cfg)
+    set_config_path(str(cfg))
 
     parser = argparse.ArgumentParser()
     parser.add_argument("--judge", dest="judge_enabled", action="store_true", default=False)
@@ -229,3 +230,35 @@ def test_apply_config_judge_cli_overrides(tmp_path, monkeypatch):
 
     assert args.judge_enabled is False  # CLI wins
     assert args.judge_confidence == 0.85  # config wins (no CLI override)
+
+
+def test_set_config_path_overrides_default(tmp_path):
+    """set_config_path() overrides the default config location."""
+    _reset_cache()
+    cfg = tmp_path / "custom.toml"
+    cfg.write_text('[api]\nbase_url = "http://custom:9999"\n')
+    set_config_path(str(cfg))
+    assert load_config("api") == {"base_url": "http://custom:9999"}
+
+
+def test_env_var_overrides_default(tmp_path, monkeypatch):
+    """$TURNSTONE_CONFIG env var overrides the default config location."""
+    _reset_cache()
+    cfg = tmp_path / "env.toml"
+    cfg.write_text('[api]\nbase_url = "http://env:7777"\n')
+    monkeypatch.setenv("TURNSTONE_CONFIG", str(cfg))
+    assert load_config("api") == {"base_url": "http://env:7777"}
+
+
+def test_set_config_path_overrides_env_var(tmp_path, monkeypatch):
+    """set_config_path() takes precedence over $TURNSTONE_CONFIG."""
+    _reset_cache()
+    env_cfg = tmp_path / "env.toml"
+    env_cfg.write_text('[api]\nbase_url = "http://env"\n')
+    monkeypatch.setenv("TURNSTONE_CONFIG", str(env_cfg))
+
+    explicit_cfg = tmp_path / "explicit.toml"
+    explicit_cfg.write_text('[api]\nbase_url = "http://explicit"\n')
+    set_config_path(str(explicit_cfg))
+
+    assert load_config("api") == {"base_url": "http://explicit"}

tests/test_console.py

@@ -3,7 +3,7 @@
 import asyncio
 import json
 import queue
-from unittest.mock import MagicMock
+from unittest.mock import MagicMock, patch
 
 import pytest
 
@@ -47,8 +47,8 @@ class MockBroker:
 # ---------------------------------------------------------------------------
 
 
-def _make_collector(broker=None, poll_interval=999, discovery_interval=999):
-    """Create a collector with long intervals so threads don't auto-fire."""
+def _make_collector(broker=None, poll_interval=0, discovery_interval=999):
+    """Create a collector with zero poll interval (no jitter delay in tests)."""
     b = broker or MockBroker()
     return ClusterCollector(
         broker=b,
@@ -264,6 +264,57 @@ class TestCollectorPolling:
         assert q.empty()
         assert len(c._nodes["node-a"].workstreams) == 0
 
+    def test_poll_401_preserves_workstreams_and_marks_unreachable(self):
+        """A 401 from the server must NOT wipe workstream data."""
+        c = _make_collector()
+        c._nodes["node-a"] = NodeSnapshot(
+            node_id="node-a",
+            server_url="http://a:8080",
+            reachable=True,
+            workstreams={"ws1": {"id": "ws1", "name": "existing", "state": "idle"}},
+        )
+
+        # Mock httpx to return 401
+        import httpx as _httpx
+
+        mock_response = _httpx.Response(
+            401,
+            json={"error": "Unauthorized"},
+            request=_httpx.Request("GET", "http://a:8080/v1/api/dashboard"),
+        )
+
+        with patch.object(c._http_client, "get", return_value=mock_response):
+            c._poll_all_nodes()
+
+        # Workstream data must be preserved, node marked unreachable
+        assert c._nodes["node-a"].reachable is False
+        assert "ws1" in c._nodes["node-a"].workstreams
+        assert c._nodes["node-a"].workstreams["ws1"]["name"] == "existing"
+
+    def test_poll_403_preserves_workstreams(self):
+        """A 403 should also preserve state and mark unreachable."""
+        c = _make_collector()
+        c._nodes["node-a"] = NodeSnapshot(
+            node_id="node-a",
+            server_url="http://a:8080",
+            reachable=True,
+            workstreams={"ws1": {"id": "ws1", "name": "keep-me", "state": "running"}},
+        )
+
+        import httpx as _httpx
+
+        mock_response = _httpx.Response(
+            403,
+            json={"error": "Forbidden"},
+            request=_httpx.Request("GET", "http://a:8080/v1/api/dashboard"),
+        )
+
+        with patch.object(c._http_client, "get", return_value=mock_response):
+            c._poll_all_nodes()
+
+        assert c._nodes["node-a"].reachable is False
+        assert "ws1" in c._nodes["node-a"].workstreams
+
 
 class TestCollectorEvents:
     """Real-time event handling from cluster channel."""
@@ -902,6 +953,8 @@ class TestConsoleWorkstreamCreation:
             ],
             2,
         )
+        # get_all_nodes delegates to get_nodes (mirrors real implementation)
+        collector.get_all_nodes.side_effect = lambda: collector.get_nodes.return_value[0]
         return collector
 
     @pytest.fixture()
@@ -1050,6 +1103,39 @@ class TestConsoleWorkstreamCreation:
         assert resp.status_code == 200
         assert resp.json()["target_node"] == "pool"
 
+    def test_create_with_resume_ws_directed(self, client_and_broker, mock_collector):
+        """resume_ws is forwarded in directed dispatch."""
+        client, broker = client_and_broker
+        resp = client.post(
+            "/v1/api/cluster/workstreams/new",
+            json={"node_id": "node-a", "resume_ws": "old-ws-id-123"},
+        )
+        assert resp.status_code == 200
+        msg = json.loads(broker.push_inbound.call_args[0][0])
+        assert msg["resume_ws"] == "old-ws-id-123"
+
+    def test_create_with_resume_ws_pool(self, client_and_broker, mock_collector):
+        """resume_ws is forwarded in pool dispatch."""
+        client, broker = client_and_broker
+        resp = client.post(
+            "/v1/api/cluster/workstreams/new",
+            json={"node_id": "pool", "resume_ws": "old-ws-id-456"},
+        )
+        assert resp.status_code == 200
+        msg = json.loads(broker.push_inbound.call_args[0][0])
+        assert msg["resume_ws"] == "old-ws-id-456"
+
+    def test_create_with_resume_ws_auto(self, client_and_broker, mock_collector):
+        """resume_ws is forwarded in auto-select dispatch."""
+        client, broker = client_and_broker
+        resp = client.post(
+            "/v1/api/cluster/workstreams/new",
+            json={"resume_ws": "old-ws-id-789"},
+        )
+        assert resp.status_code == 200
+        msg = json.loads(broker.push_inbound.call_args[0][0])
+        assert msg["resume_ws"] == "old-ws-id-789"
+
 
 # ---------------------------------------------------------------------------
 # Proxy tests
@@ -1182,47 +1268,49 @@ class TestProxyRewriting:
 class TestPickBestNode:
     """Test the _pick_best_node helper."""
 
+    @staticmethod
+    def _mock_collector(nodes: list) -> MagicMock:
+        collector = MagicMock(spec=ClusterCollector)
+        collector.get_nodes.return_value = (nodes, len(nodes))
+        collector.get_all_nodes.side_effect = lambda: collector.get_nodes.return_value[0]
+        return collector
+
     def test_picks_node_with_most_headroom(self):
         from turnstone.console.server import _pick_best_node
 
-        collector = MagicMock(spec=ClusterCollector)
-        collector.get_nodes.return_value = (
+        collector = self._mock_collector(
             [
                 {"node_id": "busy", "reachable": True, "max_ws": 10, "ws_total": 9},
                 {"node_id": "free", "reachable": True, "max_ws": 10, "ws_total": 2},
                 {"node_id": "mid", "reachable": True, "max_ws": 10, "ws_total": 5},
-            ],
-            3,
+            ]
         )
         assert _pick_best_node(collector) == "free"
 
     def test_skips_unreachable_nodes(self):
         from turnstone.console.server import _pick_best_node
 
-        collector = MagicMock(spec=ClusterCollector)
-        collector.get_nodes.return_value = (
+        collector = self._mock_collector(
             [
                 {"node_id": "down", "reachable": False, "max_ws": 10, "ws_total": 0},
                 {"node_id": "up", "reachable": True, "max_ws": 10, "ws_total": 5},
-            ],
-            2,
+            ]
         )
         assert _pick_best_node(collector) == "up"
 
     def test_returns_empty_when_no_nodes(self):
         from turnstone.console.server import _pick_best_node
 
-        collector = MagicMock(spec=ClusterCollector)
-        collector.get_nodes.return_value = ([], 0)
+        collector = self._mock_collector([])
         assert _pick_best_node(collector) == ""
 
     def test_returns_empty_when_all_unreachable(self):
         from turnstone.console.server import _pick_best_node
 
-        collector = MagicMock(spec=ClusterCollector)
-        collector.get_nodes.return_value = (
-            [{"node_id": "down", "reachable": False, "max_ws": 10, "ws_total": 0}],
-            1,
+        collector = self._mock_collector(
+            [
+                {"node_id": "down", "reachable": False, "max_ws": 10, "ws_total": 0},
+            ]
         )
         assert _pick_best_node(collector) == ""
 
@@ -1611,6 +1699,236 @@ class TestSSEProxy:
         asyncio.run(_run())
 
 
+# ---------------------------------------------------------------------------
+# Proxy auth header propagation
+# ---------------------------------------------------------------------------
+
+
+class TestProxyAuthHeaders:
+    """Verify _proxy_auth_headers mints user-scoped JWTs for proxy requests."""
+
+    SECRET = "test-secret-that-is-at-least-32-chars"
+
+    def _make_request(
+        self, *, auth_result=None, jwt_secret="", proxy_token_mgr=None, proxy_auth_token=""
+    ):
+        """Build a minimal fake request for _proxy_auth_headers."""
+
+        class _State:
+            pass
+
+        class _AppState:
+            pass
+
+        class _App:
+            state = _AppState()
+
+        class _Request:
+            state = _State()
+            app = _App()
+
+        req = _Request()
+        req.state.auth_result = auth_result
+        req.app.state.jwt_secret = jwt_secret
+        req.app.state.proxy_token_mgr = proxy_token_mgr
+        req.app.state.proxy_auth_token = proxy_auth_token
+        return req
+
+    def test_mints_user_jwt(self):
+        """Real user auth_result → JWT with correct sub, scopes, src, aud, permissions."""
+        import jwt as pyjwt
+
+        from turnstone.console.server import _proxy_auth_headers
+        from turnstone.core.auth import JWT_AUD_SERVER, AuthResult
+
+        auth = AuthResult(
+            user_id="alice",
+            scopes=frozenset({"read", "write"}),
+            token_source="jwt",
+            permissions=frozenset({"admin.users"}),
+        )
+        req = self._make_request(auth_result=auth, jwt_secret=self.SECRET)
+        headers = _proxy_auth_headers(req)
+
+        assert "Authorization" in headers
+        token = headers["Authorization"].removeprefix("Bearer ")
+        payload = pyjwt.decode(token, self.SECRET, algorithms=["HS256"], audience=JWT_AUD_SERVER)
+        assert payload["sub"] == "alice"
+        assert set(payload["scopes"].split(",")) == {"read", "write"}
+        assert payload["src"] == "console-proxy"
+        assert payload["aud"] == JWT_AUD_SERVER
+        assert payload["permissions"] == "admin.users"
+
+    def test_narrows_scopes(self):
+        """Read-only user → JWT carries only read scope, not full {read,write,approve}."""
+        import jwt as pyjwt
+
+        from turnstone.console.server import _proxy_auth_headers
+        from turnstone.core.auth import JWT_AUD_SERVER, AuthResult
+
+        auth = AuthResult(
+            user_id="viewer",
+            scopes=frozenset({"read"}),
+            token_source="jwt",
+        )
+        req = self._make_request(auth_result=auth, jwt_secret=self.SECRET)
+        headers = _proxy_auth_headers(req)
+
+        token = headers["Authorization"].removeprefix("Bearer ")
+        payload = pyjwt.decode(token, self.SECRET, algorithms=["HS256"], audience=JWT_AUD_SERVER)
+        assert payload["scopes"] == "read"
+
+    def test_short_expiry(self):
+        """Minted JWT expires in 300 seconds, not hours."""
+        import jwt as pyjwt
+
+        from turnstone.console.server import _proxy_auth_headers
+        from turnstone.core.auth import AuthResult
+
+        auth = AuthResult(
+            user_id="alice",
+            scopes=frozenset({"read"}),
+            token_source="jwt",
+        )
+        req = self._make_request(auth_result=auth, jwt_secret=self.SECRET)
+        headers = _proxy_auth_headers(req)
+
+        token = headers["Authorization"].removeprefix("Bearer ")
+        payload = pyjwt.decode(
+            token, self.SECRET, algorithms=["HS256"], options={"verify_aud": False}
+        )
+        assert payload["exp"] - payload["iat"] == 300
+
+    def test_fallback_no_user(self):
+        """No auth_result → falls back to ServiceTokenManager."""
+        from turnstone.console.server import _proxy_auth_headers
+        from turnstone.core.auth import ServiceTokenManager
+
+        mgr = ServiceTokenManager(
+            user_id="console-proxy",
+            scopes=frozenset({"read", "write", "approve"}),
+            source="console",
+            secret=self.SECRET,
+        )
+        req = self._make_request(proxy_token_mgr=mgr)
+        headers = _proxy_auth_headers(req)
+
+        assert "Authorization" in headers
+        assert headers["Authorization"] == f"Bearer {mgr.token}"
+
+    def test_fallback_no_secret(self):
+        """auth_result present but empty jwt_secret → falls back to ServiceTokenManager."""
+        from turnstone.console.server import _proxy_auth_headers
+        from turnstone.core.auth import AuthResult, ServiceTokenManager
+
+        auth = AuthResult(
+            user_id="alice",
+            scopes=frozenset({"read"}),
+            token_source="jwt",
+        )
+        mgr = ServiceTokenManager(
+            user_id="console-proxy",
+            scopes=frozenset({"read", "write", "approve"}),
+            source="console",
+            secret=self.SECRET,
+        )
+        req = self._make_request(auth_result=auth, jwt_secret="", proxy_token_mgr=mgr)
+        headers = _proxy_auth_headers(req)
+
+        # Should use ServiceTokenManager, not mint a user JWT
+        assert headers["Authorization"] == f"Bearer {mgr.token}"
+
+    def test_fallback_static_token(self):
+        """No auth_result, no ServiceTokenManager → uses static proxy_auth_token."""
+        from turnstone.console.server import _proxy_auth_headers
+
+        req = self._make_request(proxy_auth_token="static-tok-123")
+        headers = _proxy_auth_headers(req)
+
+        assert headers == {"Authorization": "Bearer static-tok-123"}
+
+
+# ---------------------------------------------------------------------------
+# Server: trusted user_id forwarding on create_workstream
+# ---------------------------------------------------------------------------
+
+
+class TestCreateWorkstreamUserIdTrust:
+    """Verify that only trusted service tokens can forward user_id in create_workstream."""
+
+    def _extract_uid(self, body: dict, auth_result) -> str:
+        """Replicate the trust check from server.py:create_workstream."""
+        auth = auth_result
+        uid: str = getattr(auth, "user_id", "") or ""
+        trusted_sources = {"bridge", "console"}
+        if (
+            body.get("user_id")
+            and isinstance(body["user_id"], str)
+            and auth is not None
+            and auth.token_source in trusted_sources
+        ):
+            uid = body["user_id"]
+        return uid
+
+    def test_bridge_can_forward_user_id(self):
+        from turnstone.core.auth import AuthResult
+
+        auth = AuthResult(
+            user_id="bridge",
+            scopes=frozenset({"approve"}),
+            token_source="bridge",
+        )
+        uid = self._extract_uid({"user_id": "real-user-abc"}, auth)
+        assert uid == "real-user-abc"
+
+    def test_console_service_can_forward_user_id(self):
+        from turnstone.core.auth import AuthResult
+
+        auth = AuthResult(
+            user_id="console",
+            scopes=frozenset({"approve"}),
+            token_source="console",
+        )
+        uid = self._extract_uid({"user_id": "real-user-abc"}, auth)
+        assert uid == "real-user-abc"
+
+    def test_console_proxy_user_cannot_override_user_id(self):
+        """End-user tokens via console-proxy must NOT override user_id."""
+        from turnstone.core.auth import AuthResult
+
+        auth = AuthResult(
+            user_id="real-user-abc",
+            scopes=frozenset({"read", "write"}),
+            token_source="console-proxy",
+        )
+        uid = self._extract_uid({"user_id": "impersonated-user"}, auth)
+        # Should use JWT identity, NOT the body override
+        assert uid == "real-user-abc"
+
+    def test_direct_user_cannot_override_user_id(self):
+        """Direct JWT login must NOT override user_id."""
+        from turnstone.core.auth import AuthResult
+
+        auth = AuthResult(
+            user_id="real-user-abc",
+            scopes=frozenset({"read", "write"}),
+            token_source="password",
+        )
+        uid = self._extract_uid({"user_id": "impersonated-user"}, auth)
+        assert uid == "real-user-abc"
+
+    def test_no_body_user_id_uses_jwt(self):
+        from turnstone.core.auth import AuthResult
+
+        auth = AuthResult(
+            user_id="bridge",
+            scopes=frozenset({"approve"}),
+            token_source="bridge",
+        )
+        uid = self._extract_uid({"name": "test-ws"}, auth)
+        assert uid == "bridge"
+
+
 # ---------------------------------------------------------------------------
 # Collector — MCP aggregation in get_overview()
 # ---------------------------------------------------------------------------

tests/test_env_scrub.py

@@ -0,0 +1,149 @@
+"""Tests for turnstone.core.env — subprocess environment scrubbing."""
+
+from __future__ import annotations
+
+import os
+from unittest.mock import patch
+
+from turnstone.core.env import _is_safe, _is_secret, scrubbed_env
+
+
+class TestIsSecret:
+    def test_explicit_scrub_list(self):
+        assert _is_secret("OPENAI_API_KEY") is True
+        assert _is_secret("ANTHROPIC_API_KEY") is True
+        assert _is_secret("TURNSTONE_JWT_SECRET") is True
+        assert _is_secret("AWS_SECRET_ACCESS_KEY") is True
+
+    def test_suffix_matching(self):
+        assert _is_secret("MY_CUSTOM_API_KEY") is True
+        assert _is_secret("DB_PASSWORD") is True
+        assert _is_secret("AUTH_TOKEN") is True
+        assert _is_secret("SERVICE_CREDENTIAL") is True
+        assert _is_secret("GCP_CREDENTIALS") is True
+
+    def test_safe_vars_not_secret(self):
+        assert _is_secret("PATH") is False
+        assert _is_secret("HOME") is False
+        assert _is_secret("LANG") is False
+
+    def test_no_false_positives_on_substring(self):
+        """Suffix matching avoids false positives like MONKEYTYPE."""
+        assert _is_secret("MONKEYTYPE") is False
+        assert _is_secret("KEYBOARD_LAYOUT") is False
+        assert _is_secret("PYTHONPATH") is False
+        assert _is_secret("EDITOR") is False
+        assert _is_secret("GOPATH") is False
+
+
+class TestIsSafe:
+    def test_safe_names(self):
+        assert _is_safe("PATH") is True
+        assert _is_safe("HOME") is True
+        assert _is_safe("TERM") is True
+        assert _is_safe("MANWIDTH") is True
+
+    def test_safe_prefixes(self):
+        assert _is_safe("LC_ALL") is True
+        assert _is_safe("LC_CTYPE") is True
+        assert _is_safe("XDG_RUNTIME_DIR") is True
+
+    def test_non_safe_names(self):
+        assert _is_safe("OPENAI_API_KEY") is False
+        assert _is_safe("CUSTOM_VAR") is False
+
+
+class TestScrubbedEnv:
+    def test_strips_api_keys(self):
+        fake_env = {
+            "PATH": "/usr/bin",
+            "HOME": "/home/user",
+            "OPENAI_API_KEY": "sk-secret",
+            "ANTHROPIC_API_KEY": "ant-secret",
+            "CUSTOM_VAR": "safe_value",
+        }
+        with patch.dict(os.environ, fake_env, clear=True):
+            result = scrubbed_env()
+
+        assert result["PATH"] == "/usr/bin"
+        assert result["HOME"] == "/home/user"
+        assert result["CUSTOM_VAR"] == "safe_value"
+        assert "OPENAI_API_KEY" not in result
+        assert "ANTHROPIC_API_KEY" not in result
+
+    def test_strips_pattern_matched_secrets(self):
+        fake_env = {
+            "PATH": "/usr/bin",
+            "MY_SERVICE_TOKEN": "tok-123",
+            "DB_PASSWORD": "pass123",
+        }
+        with patch.dict(os.environ, fake_env, clear=True):
+            result = scrubbed_env()
+
+        assert "MY_SERVICE_TOKEN" not in result
+        assert "DB_PASSWORD" not in result
+
+    def test_extra_vars_merged(self):
+        fake_env = {"PATH": "/usr/bin"}
+        with patch.dict(os.environ, fake_env, clear=True):
+            result = scrubbed_env(extra={"MANWIDTH": "80"})
+
+        assert result["MANWIDTH"] == "80"
+        assert result["PATH"] == "/usr/bin"
+
+    def test_passthrough_overrides_scrub(self):
+        fake_env = {
+            "PATH": "/usr/bin",
+            "OPENAI_API_KEY": "sk-needed",
+        }
+        with patch.dict(os.environ, fake_env, clear=True):
+            result = scrubbed_env(passthrough=["OPENAI_API_KEY"])
+
+        assert result["OPENAI_API_KEY"] == "sk-needed"
+
+    def test_preserves_locale_vars(self):
+        fake_env = {
+            "PATH": "/usr/bin",
+            "LC_ALL": "en_US.UTF-8",
+            "LC_CTYPE": "en_US.UTF-8",
+        }
+        with patch.dict(os.environ, fake_env, clear=True):
+            result = scrubbed_env()
+
+        assert result["LC_ALL"] == "en_US.UTF-8"
+        assert result["LC_CTYPE"] == "en_US.UTF-8"
+
+    def test_preserves_unknown_non_secret_vars(self):
+        fake_env = {
+            "PATH": "/usr/bin",
+            "PYTHONPATH": "/opt/lib",
+            "GOPATH": "/home/user/go",
+        }
+        with patch.dict(os.environ, fake_env, clear=True):
+            result = scrubbed_env()
+
+        assert result["PYTHONPATH"] == "/opt/lib"
+        assert result["GOPATH"] == "/home/user/go"
+
+    def test_extra_can_reintroduce_scrubbed_var(self):
+        """extra= intentionally overrides scrubbing (operator-controlled)."""
+        fake_env = {"PATH": "/usr/bin", "OPENAI_API_KEY": "sk-original"}
+        with patch.dict(os.environ, fake_env, clear=True):
+            result = scrubbed_env(extra={"OPENAI_API_KEY": "sk-injected"})
+
+        assert result["OPENAI_API_KEY"] == "sk-injected"
+
+    def test_less_prefix_does_not_leak_secrets(self):
+        """LESS pager vars are safe but LESS_SECRET_TOKEN is not."""
+        fake_env = {
+            "PATH": "/usr/bin",
+            "LESS": "-R",
+            "LESSOPEN": "| lesspipe %s",
+            "LESS_SECRET_TOKEN": "tok-secret",
+        }
+        with patch.dict(os.environ, fake_env, clear=True):
+            result = scrubbed_env()
+
+        assert result["LESS"] == "-R"
+        assert result["LESSOPEN"] == "| lesspipe %s"
+        assert "LESS_SECRET_TOKEN" not in result

tests/test_governance_storage.py

@@ -8,18 +8,8 @@ from __future__ import annotations
 
 from datetime import UTC, datetime
 
-import pytest
 import sqlalchemy as sa
 
-from turnstone.core.storage._sqlite import SQLiteBackend
-
-
-@pytest.fixture()
-def db(tmp_path):
-    """Create a fresh SQLite backend for each test."""
-    return SQLiteBackend(str(tmp_path / "test.db"))
-
-
 # ---------------------------------------------------------------------------
 # Roles
 # ---------------------------------------------------------------------------

tests/test_healthcheck.py

@@ -221,6 +221,63 @@ class TestBackendHealthMonitor:
         mon.record_failure()
         assert mon.circuit_state == CircuitState.OPEN  # type: ignore[comparison-overlap]
 
+    def test_probe_loop_autonomous_recovery(
+        self, mock_client: MagicMock, mock_metrics: MagicMock
+    ) -> None:
+        """_probe_loop transitions OPEN → HALF_OPEN → CLOSED without user requests."""
+        # Use very short intervals so the test is fast
+        mon = BackendHealthMonitor(
+            client=mock_client,
+            probe_interval=0.05,
+            probe_timeout=1.0,
+            failure_threshold=1,
+            cooldown=0.1,
+        )
+        # Trip the circuit
+        mon.record_failure()
+        assert mon.circuit_state == CircuitState.OPEN
+
+        # Backend is healthy — probe_once will succeed
+        mock_client.with_options.return_value.models.list.return_value = MagicMock()
+
+        # Start the probe loop and wait for autonomous recovery
+        mon.start()
+        try:
+            import time
+
+            deadline = time.monotonic() + 5.0
+            while mon.circuit_state != CircuitState.CLOSED and time.monotonic() < deadline:
+                time.sleep(0.05)
+            assert mon.circuit_state == CircuitState.CLOSED
+            # User requests should flow again without anyone calling acquire_request_permit
+            assert mon.acquire_request_permit() is True
+        finally:
+            mon.stop()
+            if mon._thread:
+                mon._thread.join(timeout=2.0)
+
+    def test_probe_loop_no_user_permit_during_probe(
+        self, mock_client: MagicMock, mock_metrics: MagicMock
+    ) -> None:
+        """While background probe is in HALF_OPEN, user requests are blocked."""
+        mon = BackendHealthMonitor(
+            client=mock_client,
+            probe_interval=0.05,
+            probe_timeout=1.0,
+            failure_threshold=1,
+            cooldown=0.1,
+        )
+        mon.record_failure()
+        assert mon.circuit_state == CircuitState.OPEN
+
+        # Force into HALF_OPEN as the probe loop would
+        with mon._lock:
+            mon._state = CircuitState.HALF_OPEN
+            mon._half_open_permit = False  # probe consumes it
+
+        # User requests should be blocked — only the probe gets through
+        assert mon.acquire_request_permit() is False
+
     def test_stop_thread(self, mock_client: MagicMock) -> None:
         """stop() signals the probe loop to exit."""
         mon = _make_monitor(mock_client)

tests/test_judge.py

@@ -182,7 +182,6 @@ class TestErrorHandling:
         result = judge._evaluate_single(
             _make_item(),
             [{"role": "user", "content": "test"}],
-            MagicMock(),
         )
         assert result is None
 
@@ -215,7 +214,6 @@ class TestErrorHandling:
         result = judge._evaluate_single(
             _make_item(),
             [{"role": "user", "content": "test"}],
-            MagicMock(),
         )
         assert result is None
 
@@ -259,7 +257,6 @@ class TestMultiTurnToolUse:
         verdict = judge._evaluate_single(
             _make_item(),
             [{"role": "user", "content": "test"}],
-            MagicMock(),
         )
         assert verdict is not None
         assert verdict.tier == "llm"
@@ -305,7 +302,6 @@ class TestMultiTurnToolUse:
         judge._evaluate_single(
             _make_item(),
             [{"role": "user", "content": "test"}],
-            MagicMock(),
         )
         # Should have called create_completion exactly _JUDGE_MAX_TURNS times
         assert provider.create_completion.call_count == 5

tests/test_judge_storage.py

@@ -4,16 +4,6 @@ from __future__ import annotations
 
 from datetime import UTC, datetime, timedelta
 
-import pytest
-
-from turnstone.core.storage._sqlite import SQLiteBackend
-
-
-@pytest.fixture()
-def db(tmp_path):
-    """Fresh SQLite backend for each test."""
-    return SQLiteBackend(str(tmp_path / "test.db"))
-
 
 def _make_verdict_kwargs(**overrides):
     """Build default kwargs for create_intent_verdict."""

tests/test_load_skill.py

@@ -1,4 +1,4 @@
-"""Tests for the load_skill built-in tool."""
+"""Tests for the skill built-in tool."""
 
 from __future__ import annotations
 
@@ -9,25 +9,25 @@ from turnstone.core.tools import BUILTIN_TOOL_NAMES, PRIMARY_KEY_MAP
 
 
 class TestToolRegistration:
-    """Verify load_skill is registered correctly."""
+    """Verify skill is registered correctly."""
 
     def test_in_builtin_tool_names(self) -> None:
-        assert "load_skill" in BUILTIN_TOOL_NAMES
+        assert "skill" in BUILTIN_TOOL_NAMES
 
     def test_not_agent_tool(self) -> None:
         from turnstone.core.tools import AGENT_TOOLS
 
         names = {t["function"]["name"] for t in AGENT_TOOLS}
-        assert "load_skill" not in names
+        assert "skill" not in names
 
     def test_not_task_agent_tool(self) -> None:
         from turnstone.core.tools import TASK_AGENT_TOOLS
 
         names = {t["function"]["name"] for t in TASK_AGENT_TOOLS}
-        assert "load_skill" not in names
+        assert "skill" not in names
 
     def test_has_primary_key(self) -> None:
-        assert PRIMARY_KEY_MAP.get("load_skill") == "name"
+        assert PRIMARY_KEY_MAP.get("skill") == "name"
 
 
 # ---------------------------------------------------------------------------
@@ -82,12 +82,12 @@ def _make_session(skills: list[dict[str, Any]] | None = None):
 
 
 class TestPrepareLoadSkill:
-    """Test _prepare_load_skill validation and item dict shape."""
+    """Test _prepare_skill validation and item dict shape."""
 
     def test_load_valid(self) -> None:
         session, _, _ = _make_session()
-        item = session._prepare_load_skill("call-1", {"action": "load", "name": "code-review"})
-        assert item["func_name"] == "load_skill"
+        item = session._prepare_skill("call-1", {"action": "load", "name": "code-review"})
+        assert item["func_name"] == "skill"
         assert item["action"] == "load"
         assert item["name"] == "code-review"
         assert item["needs_approval"] is True
@@ -96,19 +96,19 @@ class TestPrepareLoadSkill:
 
     def test_load_missing_name(self) -> None:
         session, _, _ = _make_session()
-        item = session._prepare_load_skill("call-1", {"action": "load"})
+        item = session._prepare_skill("call-1", {"action": "load"})
         assert "error" in item
         assert "name" in item["error"].lower()
         assert item["needs_approval"] is False
 
     def test_load_empty_name(self) -> None:
         session, _, _ = _make_session()
-        item = session._prepare_load_skill("call-1", {"action": "load", "name": ""})
+        item = session._prepare_skill("call-1", {"action": "load", "name": ""})
         assert "error" in item
 
     def test_search_with_query(self) -> None:
         session, _, _ = _make_session()
-        item = session._prepare_load_skill("call-1", {"action": "search", "query": "code review"})
+        item = session._prepare_skill("call-1", {"action": "search", "query": "code review"})
         assert item["action"] == "search"
         assert item["query"] == "code review"
         assert item["needs_approval"] is False
@@ -116,30 +116,30 @@ class TestPrepareLoadSkill:
 
     def test_search_without_query(self) -> None:
         session, _, _ = _make_session()
-        item = session._prepare_load_skill("call-1", {"action": "search"})
+        item = session._prepare_skill("call-1", {"action": "search"})
         assert item["action"] == "search"
         assert item["query"] == ""
         assert item["needs_approval"] is False
 
     def test_invalid_action(self) -> None:
         session, _, _ = _make_session()
-        item = session._prepare_load_skill("call-1", {"action": "delete"})
+        item = session._prepare_skill("call-1", {"action": "delete"})
         assert "error" in item
         assert "delete" in item["error"]
 
     def test_empty_action(self) -> None:
         session, _, _ = _make_session()
-        item = session._prepare_load_skill("call-1", {"action": ""})
+        item = session._prepare_skill("call-1", {"action": ""})
         assert "error" in item
 
     def test_header_for_load(self) -> None:
         session, _, _ = _make_session()
-        item = session._prepare_load_skill("call-1", {"action": "load", "name": "my-skill"})
+        item = session._prepare_skill("call-1", {"action": "load", "name": "my-skill"})
         assert "my-skill" in item["header"]
 
     def test_header_for_search(self) -> None:
         session, _, _ = _make_session()
-        item = session._prepare_load_skill("call-1", {"action": "search", "query": "testing"})
+        item = session._prepare_skill("call-1", {"action": "search", "query": "testing"})
         assert "testing" in item["header"]
 
 
@@ -149,7 +149,7 @@ class TestPrepareLoadSkill:
 
 
 class TestExecLoadSkill:
-    """Test _exec_load_skill execution logic."""
+    """Test _exec_skill execution logic."""
 
     def test_load_existing_skill(self) -> None:
         skills = [
@@ -164,8 +164,8 @@ class TestExecLoadSkill:
         session, _, fake_get = _make_session(skills)
 
         with patch("turnstone.core.session.get_skill_by_name", side_effect=fake_get):
-            item = session._prepare_load_skill("call-1", {"action": "load", "name": "code-review"})
-            call_id, result = session._exec_load_skill(item)
+            item = session._prepare_skill("call-1", {"action": "load", "name": "code-review"})
+            call_id, result = session._exec_skill(item)
 
         assert call_id == "call-1"
         assert "code-review" in result
@@ -177,8 +177,8 @@ class TestExecLoadSkill:
         session, _, fake_get = _make_session([])
 
         with patch("turnstone.core.session.get_skill_by_name", side_effect=fake_get):
-            item = session._prepare_load_skill("call-1", {"action": "load", "name": "nope"})
-            call_id, result = session._exec_load_skill(item)
+            item = session._prepare_skill("call-1", {"action": "load", "name": "nope"})
+            call_id, result = session._exec_skill(item)
 
         assert "not found" in result.lower()
         assert session._set_skill_called == []
@@ -188,8 +188,8 @@ class TestExecLoadSkill:
         session, _, fake_get = _make_session(skills)
 
         with patch("turnstone.core.session.get_skill_by_name", side_effect=fake_get):
-            item = session._prepare_load_skill("call-1", {"action": "load", "name": "test"})
-            session._exec_load_skill(item)
+            item = session._prepare_skill("call-1", {"action": "load", "name": "test"})
+            session._exec_skill(item)
 
         session.ui.on_tool_result.assert_called_once()
 
@@ -216,10 +216,10 @@ class TestExecLoadSkill:
         mock_storage.list_prompt_templates.return_value = skills
 
         session, _, _ = _make_session()
-        item = session._prepare_load_skill("call-1", {"action": "search", "query": "code"})
+        item = session._prepare_skill("call-1", {"action": "search", "query": "code"})
 
         with patch("turnstone.core.storage._registry.get_storage", return_value=mock_storage):
-            call_id, result = session._exec_load_skill(item)
+            call_id, result = session._exec_skill(item)
 
         assert "code-review" in result
         # docs-writer shouldn't match "code" query
@@ -241,10 +241,10 @@ class TestExecLoadSkill:
         mock_storage.list_prompt_templates.return_value = skills
 
         session, _, _ = _make_session()
-        item = session._prepare_load_skill("call-1", {"action": "search"})
+        item = session._prepare_skill("call-1", {"action": "search"})
 
         with patch("turnstone.core.storage._registry.get_storage", return_value=mock_storage):
-            call_id, result = session._exec_load_skill(item)
+            call_id, result = session._exec_skill(item)
 
         # Should be limited to 10
         assert result.count("skill-") == 10
@@ -254,10 +254,10 @@ class TestExecLoadSkill:
         mock_storage.list_prompt_templates.return_value = []
 
         session, _, _ = _make_session()
-        item = session._prepare_load_skill("call-1", {"action": "search", "query": "nonexistent"})
+        item = session._prepare_skill("call-1", {"action": "search", "query": "nonexistent"})
 
         with patch("turnstone.core.storage._registry.get_storage", return_value=mock_storage):
-            call_id, result = session._exec_load_skill(item)
+            call_id, result = session._exec_skill(item)
 
         assert "no skills found" in result.lower()
 
@@ -276,21 +276,21 @@ class TestExecLoadSkill:
         mock_storage.list_prompt_templates.return_value = skills
 
         session, _, _ = _make_session()
-        item = session._prepare_load_skill("call-1", {"action": "search", "query": "risky"})
+        item = session._prepare_skill("call-1", {"action": "search", "query": "risky"})
 
         with patch("turnstone.core.storage._registry.get_storage", return_value=mock_storage):
-            call_id, result = session._exec_load_skill(item)
+            call_id, result = session._exec_skill(item)
 
         assert "high" in result
 
     def test_search_storage_failure_returns_empty(self) -> None:
         session, _, _ = _make_session()
-        item = session._prepare_load_skill("call-1", {"action": "search", "query": "test"})
+        item = session._prepare_skill("call-1", {"action": "search", "query": "test"})
 
         with patch(
             "turnstone.core.storage._registry.get_storage", side_effect=RuntimeError("no storage")
         ):
-            call_id, result = session._exec_load_skill(item)
+            call_id, result = session._exec_skill(item)
 
         assert "no skills found" in result.lower()
 
@@ -307,10 +307,8 @@ class TestExecLoadSkill:
         session, _, fake_get = _make_session(skills)
 
         with patch("turnstone.core.session.get_skill_by_name", side_effect=fake_get):
-            item = session._prepare_load_skill(
-                "call-1", {"action": "load", "name": "disabled-skill"}
-            )
-            call_id, result = session._exec_load_skill(item)
+            item = session._prepare_skill("call-1", {"action": "load", "name": "disabled-skill"})
+            call_id, result = session._exec_skill(item)
 
         assert "not found" in result.lower()
         assert session._set_skill_called == []
@@ -321,8 +319,8 @@ class TestExecLoadSkill:
         session._skill_name = "active"
 
         with patch("turnstone.core.session.get_skill_by_name", side_effect=fake_get):
-            item = session._prepare_load_skill("call-1", {"action": "load", "name": "active"})
-            call_id, result = session._exec_load_skill(item)
+            item = session._prepare_skill("call-1", {"action": "load", "name": "active"})
+            call_id, result = session._exec_skill(item)
 
         assert "already active" in result.lower()
         assert session._set_skill_called == []
@@ -352,10 +350,10 @@ class TestExecLoadSkill:
         mock_storage.list_prompt_templates.return_value = skills
 
         session, _, _ = _make_session()
-        item = session._prepare_load_skill("call-1", {"action": "search"})
+        item = session._prepare_skill("call-1", {"action": "search"})
 
         with patch("turnstone.core.storage._registry.get_storage", return_value=mock_storage):
-            call_id, result = session._exec_load_skill(item)
+            call_id, result = session._exec_skill(item)
 
         assert "enabled-skill" in result
         assert "disabled-skill" not in result
@@ -375,14 +373,113 @@ class TestExecLoadSkill:
         mock_storage.list_prompt_templates.return_value = skills
 
         session, _, _ = _make_session()
-        item = session._prepare_load_skill("call-1", {"action": "search", "query": "code review"})
+        item = session._prepare_skill("call-1", {"action": "search", "query": "code review"})
 
         with patch("turnstone.core.storage._registry.get_storage", return_value=mock_storage):
-            call_id, result = session._exec_load_skill(item)
+            call_id, result = session._exec_skill(item)
 
         assert "code-review" in result
 
     def test_preparer_load_has_approval_label(self) -> None:
         session, _, _ = _make_session()
-        item = session._prepare_load_skill("call-1", {"action": "load", "name": "my-skill"})
-        assert item["approval_label"] == "load_skill__my-skill"
+        item = session._prepare_skill("call-1", {"action": "load", "name": "my-skill"})
+        assert item["approval_label"] == "skill__my-skill"
+
+
+# ---------------------------------------------------------------------------
+# Tests: Skill Catalog Disclosure (Agent Skills standard compliance)
+# ---------------------------------------------------------------------------
+
+
+class TestSkillCatalogDisclosure:
+    """Verify <available-skills> catalog appears in system messages."""
+
+    def _build_session_with_system_messages(
+        self,
+        search_skills: list[dict[str, Any]] | None = None,
+    ) -> Any:
+        """Build a session and call _init_system_messages to get dev_parts."""
+        from turnstone.core.session import ChatSession
+
+        session = ChatSession.__new__(ChatSession)
+        ui = MagicMock()
+        session.ui = ui
+        session.model = "test-model"
+        session._ws_id = "ws-test"
+        session._node_id = "node-1"
+        session._skill_name = None
+        session._skill_content = None
+        session._skill_resources = {}
+        session._applied_skill_content = None
+        session.context_window = 128000
+        session.messages = []
+        session._config = {}
+        session.creative_mode = False
+        session.instructions = ""
+        session.system_messages = []
+        session._agent_system_messages = []
+        session.reasoning_effort = "medium"
+        session._pending_nudge = []
+        session._tool_search = None
+        session._mcp_client = None
+        session._notify_on_complete = "{}"
+
+        # Memory stubs
+        session._memory_config = MagicMock()
+        session._memory_config.fetch_limit = 0
+        session._user_id = ""
+
+        with (
+            patch(
+                "turnstone.core.session.list_skills_by_activation",
+                return_value=search_skills or [],
+            ),
+            patch.object(session, "_get_visible_memories", return_value=[]),
+        ):
+            session._init_system_messages()
+
+        return session
+
+    def test_catalog_present_with_search_skills(self) -> None:
+        skills = [
+            {"name": "pdf-processing", "description": "Extract PDF text and forms."},
+            {"name": "data-analysis", "description": "Analyze datasets."},
+        ]
+        session = self._build_session_with_system_messages(search_skills=skills)
+        content = session.system_messages[0]["content"]
+        assert "<available-skills>" in content
+        assert "pdf-processing" in content
+        assert "data-analysis" in content
+        assert "</available-skills>" in content
+
+    def test_catalog_omitted_when_no_search_skills(self) -> None:
+        session = self._build_session_with_system_messages(search_skills=[])
+        content = session.system_messages[0]["content"]
+        assert "<available-skills>" not in content
+
+    def test_catalog_capped_at_30(self) -> None:
+        skills = [{"name": f"skill-{i:03d}", "description": f"Desc {i}"} for i in range(50)]
+        session = self._build_session_with_system_messages(search_skills=skills)
+        content = session.system_messages[0]["content"]
+        # Should include first 30, not all 50
+        assert "skill-029" in content
+        assert "skill-030" not in content
+
+    def test_catalog_escapes_html(self) -> None:
+        skills = [
+            {"name": "xss-test", "description": "Handle <script> & 'quotes'."},
+        ]
+        session = self._build_session_with_system_messages(search_skills=skills)
+        content = session.system_messages[0]["content"]
+        assert "&lt;script&gt;" in content
+        assert "<script>" not in content.replace("<available-skills>", "").replace(
+            "</available-skills>", ""
+        ).replace("<skill>", "").replace("</skill>", "").replace("<name>", "").replace(
+            "</name>", ""
+        ).replace("<description>", "").replace("</description>", "")
+
+    def test_catalog_includes_hint(self) -> None:
+        skills = [{"name": "test", "description": "Test skill."}]
+        session = self._build_session_with_system_messages(search_skills=skills)
+        content = session.system_messages[0]["content"]
+        assert "/skill" in content

tests/test_mcp_admin_api.py

@@ -26,6 +26,7 @@ from turnstone.console.server import (
     admin_get_mcp_server,
     admin_import_mcp_config,
     admin_list_mcp_servers,
+    admin_mcp_reload,
     admin_update_mcp_server,
 )
 from turnstone.core.auth import AuthResult
@@ -96,6 +97,11 @@ _ROUTES = [
                 admin_import_mcp_config,
                 methods=["POST"],
             ),
+            Route(
+                "/api/admin/mcp-servers/reload",
+                admin_mcp_reload,
+                methods=["POST"],
+            ),
             Route(
                 "/api/admin/mcp-servers/{server_id}",
                 admin_get_mcp_server,
@@ -115,6 +121,21 @@ _ROUTES = [
 ]
 
 
+def _routes_with_internal() -> list[Mount]:
+    """Routes including the node-side internal endpoint (lazy-imported)."""
+    from turnstone.server import internal_mcp_reload
+
+    return [
+        Mount(
+            "/v1",
+            routes=[
+                *_ROUTES[0].routes,  # type: ignore[union-attr]
+                Route("/api/_internal/mcp-reload", internal_mcp_reload, methods=["POST"]),
+            ],
+        ),
+    ]
+
+
 @pytest.fixture
 def storage(tmp_path):
     return SQLiteBackend(str(tmp_path / "test.db"))
@@ -550,8 +571,11 @@ def _fake_request(*nodes: dict[str, Any], proxy_client: Any = None) -> MagicMock
     """Build a minimal mock request with collector and proxy_client."""
     collector = MagicMock()
     collector.get_nodes.return_value = (list(nodes), len(nodes))
+    collector.get_all_nodes.side_effect = lambda: collector.get_nodes.return_value[0]
     req = MagicMock()
+    req.state.auth_result = None
     req.app.state.collector = collector
+    req.app.state.jwt_secret = ""
     req.app.state.proxy_client = proxy_client or AsyncMock()
     req.app.state.proxy_token_mgr = None
     req.app.state.proxy_auth_token = "tok"
@@ -693,3 +717,175 @@ class TestNotifyNodesMcpReload:
         result = await _notify_nodes_mcp_reload(req)
         assert result["n1"] == {"reloaded": 2}
         assert "error" in result["n2"]
+
+
+# ---------------------------------------------------------------------------
+# Console reload endpoint: POST /v1/api/admin/mcp-servers/reload
+# ---------------------------------------------------------------------------
+
+
+class TestAdminMcpReloadEndpoint:
+    """HTTP-level tests for the console reload endpoint."""
+
+    def test_reload_success(self, client: TestClient) -> None:
+        """Reload endpoint returns status ok and fan-out results."""
+        with patch(
+            "turnstone.console.server._notify_nodes_mcp_reload",
+            new_callable=AsyncMock,
+            return_value={"n1": {"reloaded": 3}},
+        ):
+            r = client.post("/v1/api/admin/mcp-servers/reload")
+        assert r.status_code == 200
+        data = r.json()
+        assert data["status"] == "ok"
+        assert data["results"] == {"n1": {"reloaded": 3}}
+
+    def test_reload_empty_cluster(self, client: TestClient) -> None:
+        """Reload with no nodes returns empty results."""
+        with patch(
+            "turnstone.console.server._notify_nodes_mcp_reload",
+            new_callable=AsyncMock,
+            return_value={},
+        ):
+            r = client.post("/v1/api/admin/mcp-servers/reload")
+        assert r.status_code == 200
+        data = r.json()
+        assert data["status"] == "ok"
+        assert data["results"] == {}
+
+    def test_reload_permission_denied(self, client_no_perm: TestClient) -> None:
+        """Reload without admin.mcp permission is rejected."""
+        r = client_no_perm.post("/v1/api/admin/mcp-servers/reload")
+        assert r.status_code == 403
+        assert "admin.mcp" in r.json()["error"]
+
+    def test_reload_no_storage(self) -> None:
+        """Reload returns 503 when auth_storage is not available."""
+        app = Starlette(
+            routes=_ROUTES,
+            middleware=[Middleware(_InjectAuthMiddleware)],
+        )
+        # Deliberately omit app.state.auth_storage
+        no_storage_client = TestClient(app, raise_server_exceptions=False)
+        r = no_storage_client.post("/v1/api/admin/mcp-servers/reload")
+        assert r.status_code == 503
+
+    def test_reload_mixed_node_results(self, client: TestClient) -> None:
+        """Reload propagates per-node errors in results."""
+        with patch(
+            "turnstone.console.server._notify_nodes_mcp_reload",
+            new_callable=AsyncMock,
+            return_value={
+                "n1": {"reloaded": 2},
+                "n2": {"error": "Connection refused"},
+            },
+        ):
+            r = client.post("/v1/api/admin/mcp-servers/reload")
+        assert r.status_code == 200
+        data = r.json()
+        assert data["results"]["n1"] == {"reloaded": 2}
+        assert "error" in data["results"]["n2"]
+
+
+# ---------------------------------------------------------------------------
+# Node reload endpoint: POST /v1/api/_internal/mcp-reload
+# ---------------------------------------------------------------------------
+
+
+class TestInternalMcpReloadEndpoint:
+    """HTTP-level tests for the node-side MCP reload endpoint."""
+
+    @pytest.fixture()
+    def node_client(self, storage: SQLiteBackend) -> TestClient:
+        """TestClient with an MCP client manager on app.state."""
+        app = Starlette(
+            routes=_routes_with_internal(),
+            middleware=[Middleware(_InjectAuthMiddleware)],
+        )
+        app.state.auth_storage = storage
+        mgr = MagicMock()
+        mgr.reconcile_sync.return_value = {
+            "added": ["new-srv"],
+            "removed": [],
+            "updated": [],
+        }
+        app.state.mcp_client = mgr
+        return TestClient(app, raise_server_exceptions=False)
+
+    def test_reload_calls_reconcile(self, node_client: TestClient, storage: SQLiteBackend) -> None:
+        """Reload endpoint calls reconcile_sync and returns its result."""
+        with patch("turnstone.core.storage._registry.get_storage", return_value=storage):
+            r = node_client.post("/v1/api/_internal/mcp-reload")
+        assert r.status_code == 200
+        data = r.json()
+        assert data["status"] == "ok"
+        assert data["added"] == ["new-srv"]
+        assert data["removed"] == []
+        assert data["updated"] == []
+
+    def test_reload_passes_storage_to_reconcile(
+        self,
+        storage: SQLiteBackend,
+    ) -> None:
+        """Verify reconcile_sync receives the storage backend."""
+        app = Starlette(
+            routes=_routes_with_internal(),
+            middleware=[Middleware(_InjectAuthMiddleware)],
+        )
+        app.state.auth_storage = storage
+        mgr = MagicMock()
+        mgr.reconcile_sync.return_value = {"added": [], "removed": [], "updated": []}
+        app.state.mcp_client = mgr
+        c = TestClient(app, raise_server_exceptions=False)
+        with patch("turnstone.core.storage._registry.get_storage", return_value=storage):
+            r = c.post("/v1/api/_internal/mcp-reload")
+        assert r.status_code == 200
+        mgr.reconcile_sync.assert_called_once_with(storage)
+
+    def test_reload_creates_manager_when_missing(self, storage: SQLiteBackend) -> None:
+        """When mcp_client is absent, a new MCPClientManager is created."""
+        app = Starlette(
+            routes=_routes_with_internal(),
+            middleware=[Middleware(_InjectAuthMiddleware)],
+        )
+        app.state.auth_storage = storage
+        # No mcp_client on app.state
+        c = TestClient(app, raise_server_exceptions=False)
+        with (
+            patch("turnstone.core.storage._registry.get_storage", return_value=storage),
+            patch("turnstone.core.mcp_client.MCPClientManager") as mock_cls,
+        ):
+            mock_mgr = MagicMock()
+            mock_mgr.reconcile_sync.return_value = {
+                "added": [],
+                "removed": [],
+                "updated": [],
+            }
+            mock_cls.return_value = mock_mgr
+            r = c.post("/v1/api/_internal/mcp-reload")
+        assert r.status_code == 200
+        mock_cls.assert_called_once_with({})
+        mock_mgr.start.assert_called_once()
+        mock_mgr.reconcile_sync.assert_called_once_with(storage)
+
+    def test_reload_reconcile_result_in_response(self, storage: SQLiteBackend) -> None:
+        """Full reconcile result fields (added/removed/updated) appear in JSON."""
+        app = Starlette(
+            routes=_routes_with_internal(),
+            middleware=[Middleware(_InjectAuthMiddleware)],
+        )
+        app.state.auth_storage = storage
+        mgr = MagicMock()
+        mgr.reconcile_sync.return_value = {
+            "added": ["a"],
+            "removed": ["b"],
+            "updated": ["c"],
+        }
+        app.state.mcp_client = mgr
+        c = TestClient(app, raise_server_exceptions=False)
+        with patch("turnstone.core.storage._registry.get_storage", return_value=storage):
+            r = c.post("/v1/api/_internal/mcp-reload")
+        data = r.json()
+        assert data["added"] == ["a"]
+        assert data["removed"] == ["b"]
+        assert data["updated"] == ["c"]

tests/test_mcp_registry.py

@@ -398,6 +398,72 @@ class TestResolveInstallConfig:
         config = resolve_install_config(server, "remote", 0)
         assert config["url"] == "https://us-east.example.com/mcp"
 
+    def test_remote_variable_substitution_invalid_scheme(self) -> None:
+        server = RegistryServer(
+            name="io.example/test",
+            version="1.0.0",
+            remotes=[
+                RegistryRemote(
+                    type="streamable-http",
+                    url="{scheme}://evil.example.com/mcp",
+                    variables={
+                        "scheme": RegistryRemoteVariable(is_required=True),
+                    },
+                )
+            ],
+        )
+        with pytest.raises(MCPRegistryError, match="Invalid URL scheme"):
+            resolve_install_config(server, "remote", 0, variables={"scheme": "file"})
+
+    def test_remote_variable_substitution_preserves_valid_scheme(self) -> None:
+        server = RegistryServer(
+            name="io.example/test",
+            version="1.0.0",
+            remotes=[
+                RegistryRemote(
+                    type="streamable-http",
+                    url="https://{host}.example.com/mcp",
+                    variables={
+                        "host": RegistryRemoteVariable(is_required=True),
+                    },
+                )
+            ],
+        )
+        config = resolve_install_config(server, "remote", 0, variables={"host": "api"})
+        assert config["url"] == "https://api.example.com/mcp"
+
+    def test_remote_variable_substitution_missing_hostname(self) -> None:
+        """URL like https:///mcp has valid scheme but no hostname."""
+        server = RegistryServer(
+            name="io.example/test",
+            version="1.0.0",
+            remotes=[
+                RegistryRemote(
+                    type="streamable-http",
+                    url="https:///mcp",
+                )
+            ],
+        )
+        with pytest.raises(MCPRegistryError, match="hostname is missing"):
+            resolve_install_config(server, "remote", 0)
+
+    def test_remote_variable_substitution_embedded_credentials(self) -> None:
+        server = RegistryServer(
+            name="io.example/test",
+            version="1.0.0",
+            remotes=[
+                RegistryRemote(
+                    type="streamable-http",
+                    url="https://{creds}@example.com/mcp",
+                    variables={
+                        "creds": RegistryRemoteVariable(is_required=True),
+                    },
+                )
+            ],
+        )
+        with pytest.raises(MCPRegistryError, match="embedded credentials"):
+            resolve_install_config(server, "remote", 0, variables={"creds": "user:pass"})
+
     def test_remote_no_remotes(self) -> None:
         server = RegistryServer(name="io.example/test", version="1.0.0")
         with pytest.raises(MCPRegistryError, match="no remote"):

tests/test_mcp_registry_api.py

@@ -4,7 +4,7 @@ from __future__ import annotations
 
 import json
 from typing import TYPE_CHECKING, Any
-from unittest.mock import AsyncMock, patch
+from unittest.mock import AsyncMock, MagicMock, patch
 
 import pytest
 from starlette.applications import Starlette
@@ -18,11 +18,13 @@ if TYPE_CHECKING:
     from starlette.responses import Response
 
 from turnstone.console.server import (
+    _get_registry_url,
     admin_registry_install,
     admin_registry_search,
 )
 from turnstone.core.auth import AuthResult
 from turnstone.core.mcp_registry import (
+    DEFAULT_REGISTRY_URL,
     MCPRegistryError,
     RegistryPackage,
     RegistryRemote,
@@ -362,7 +364,7 @@ class TestRegistryInstall:
     def test_install_max_servers(self, client: TestClient, storage: SQLiteBackend) -> None:
         import uuid
 
-        for i in range(50):
+        for i in range(200):
             storage.create_mcp_server(
                 server_id=uuid.uuid4().hex,
                 name=f"server-{i}",
@@ -549,3 +551,73 @@ class TestRegistryInstall:
 
         assert resp.status_code == 409
         assert "custom 'name'" in resp.json()["error"]
+
+
+# ---------------------------------------------------------------------------
+# _get_registry_url fallback chain tests
+# ---------------------------------------------------------------------------
+
+
+def _mock_request(storage: Any = None, config_store: Any = None) -> MagicMock:
+    """Build a mock Request with app.state.auth_storage and app.state.config_store."""
+    request = MagicMock()
+    request.app.state.auth_storage = storage
+    request.app.state.config_store = config_store
+    return request
+
+
+class TestGetRegistryUrl:
+    """Verify three-tier URL resolution: DB setting -> config.toml -> default."""
+
+    def test_returns_db_setting_when_available(self) -> None:
+        config_store = MagicMock()
+        config_store.get.return_value = "https://custom.registry.example.com"
+        request = _mock_request(config_store=config_store)
+
+        with patch("turnstone.core.config.load_config", return_value={}):
+            url = _get_registry_url(request)
+
+        assert url == "https://custom.registry.example.com"
+        config_store.get.assert_called_once_with("mcp.registry_url")
+
+    def test_falls_back_to_config_when_config_store_returns_empty(self) -> None:
+        config_store = MagicMock()
+        config_store.get.return_value = ""
+        request = _mock_request(config_store=config_store)
+
+        with patch(
+            "turnstone.core.config.load_config",
+            return_value={"registry_url": "https://config.registry.example.com"},
+        ):
+            url = _get_registry_url(request)
+
+        assert url == "https://config.registry.example.com"
+
+    def test_falls_back_to_config_when_no_config_store(self) -> None:
+        request = _mock_request()
+
+        with patch(
+            "turnstone.core.config.load_config",
+            return_value={"registry_url": "https://config.registry.example.com"},
+        ):
+            url = _get_registry_url(request)
+
+        assert url == "https://config.registry.example.com"
+
+    def test_falls_back_to_default_when_both_unavailable(self) -> None:
+        config_store = MagicMock()
+        config_store.get.return_value = ""
+        request = _mock_request(config_store=config_store)
+
+        with patch("turnstone.core.config.load_config", return_value={}):
+            url = _get_registry_url(request)
+
+        assert url == DEFAULT_REGISTRY_URL
+
+    def test_falls_back_to_default_when_no_config_store_or_config(self) -> None:
+        request = _mock_request()
+
+        with patch("turnstone.core.config.load_config", return_value={}):
+            url = _get_registry_url(request)
+
+        assert url == DEFAULT_REGISTRY_URL

tests/test_mcp_server_storage.py

@@ -3,16 +3,10 @@
 from __future__ import annotations
 
 import uuid
+from typing import TYPE_CHECKING
 
-import pytest
-
-from turnstone.core.storage._sqlite import SQLiteBackend
-
-
-@pytest.fixture
-def db(tmp_path):
-    """Fresh SQLite backend for each test."""
-    return SQLiteBackend(str(tmp_path / "test.db"))
+if TYPE_CHECKING:
+    from turnstone.core.storage._sqlite import SQLiteBackend
 
 
 def _make_id() -> str:

tests/test_model_registry.py

@@ -10,6 +10,7 @@ import pytest
 from turnstone.core.model_registry import (
     ModelConfig,
     ModelRegistry,
+    detect_model,
     load_model_registry,
 )
 
@@ -590,3 +591,36 @@ class TestProtocolModel:
         assert isinstance(restored, CreateWorkstreamMessage)
         assert restored.model == "local"
         assert restored.name == "ws1"
+
+
+# ---------------------------------------------------------------------------
+# detect_model — startup timeout
+# ---------------------------------------------------------------------------
+
+
+class TestDetectModelTimeout:
+    def test_uses_short_timeout_and_no_retries(self) -> None:
+        """detect_model() uses with_options(timeout=10, max_retries=0)."""
+        mock_model = MagicMock()
+        mock_model.id = "test-model"
+        mock_model.owned_by = "test"
+
+        fast_client = MagicMock()
+        fast_client.models.list.return_value = MagicMock(data=[mock_model])
+
+        client = MagicMock()
+        client.with_options.return_value = fast_client
+
+        result = detect_model(client, provider="openai")
+        client.with_options.assert_called_once_with(timeout=10.0, max_retries=0)
+        fast_client.models.list.assert_called_once()
+        assert result[0] == "test-model"
+
+    def test_connection_error_non_fatal(self) -> None:
+        """detect_model(fatal=False) returns (None, None) on connection error."""
+        client = MagicMock()
+        client.with_options.return_value = client
+        client.models.list.side_effect = OSError("Connection refused")
+
+        result = detect_model(client, provider="openai", fatal=False)
+        assert result == (None, None)

tests/test_oidc.py

@@ -22,6 +22,7 @@ from turnstone.core.oidc import (
     load_oidc_config,
     provision_oidc_user,
     validate_id_token,
+    validate_issuer_url,
 )
 
 # ---------------------------------------------------------------------------
@@ -293,6 +294,162 @@ class TestLoadOIDCConfig:
         assert cfg.redirect_base == "http://localhost:8000"
 
 
+# ---------------------------------------------------------------------------
+# SSRF Validation
+# ---------------------------------------------------------------------------
+
+
+class TestValidateIssuerURL:
+    """Tests for ``validate_issuer_url`` SSRF protection."""
+
+    def test_valid_https_url(self):
+        """Public HTTPS issuer URL passes validation."""
+        # Should not raise -- mock DNS to return a public IP.
+        with patch(
+            "socket.getaddrinfo",
+            return_value=[
+                (2, 1, 6, "", ("93.184.216.34", 0)),
+            ],
+        ):
+            validate_issuer_url("https://idp.example.com")
+
+    def test_rejects_http_non_localhost(self):
+        """HTTP is rejected for non-localhost hosts."""
+        with pytest.raises(OIDCError, match="must use HTTPS"):
+            validate_issuer_url("http://idp.example.com")
+
+    def test_allows_http_localhost(self):
+        """HTTP is allowed for localhost (development)."""
+        with patch(
+            "socket.getaddrinfo",
+            return_value=[
+                (2, 1, 6, "", ("127.0.0.1", 0)),
+            ],
+        ):
+            validate_issuer_url("http://localhost:8080")
+
+    def test_allows_http_localhost_subdomain(self):
+        """HTTP is allowed for *.localhost subdomains."""
+        with patch(
+            "socket.getaddrinfo",
+            return_value=[
+                (2, 1, 6, "", ("127.0.0.1", 0)),
+            ],
+        ):
+            validate_issuer_url("http://keycloak.localhost:8080")
+
+    def test_rejects_embedded_credentials(self):
+        """URLs with userinfo (user:pass@host) are rejected."""
+        with pytest.raises(OIDCError, match="embedded credentials"):
+            validate_issuer_url("https://admin:secret@idp.example.com")
+
+    def test_rejects_username_only(self):
+        """URLs with just a username are rejected."""
+        with pytest.raises(OIDCError, match="embedded credentials"):
+            validate_issuer_url("https://admin@idp.example.com")
+
+    def test_rejects_no_hostname(self):
+        """URLs without a hostname are rejected."""
+        with pytest.raises(OIDCError, match="no hostname"):
+            validate_issuer_url("https://")
+
+    def test_rejects_private_10_range(self):
+        """Hostnames resolving to 10.x.x.x are rejected."""
+        with (
+            patch("socket.getaddrinfo", return_value=[(2, 1, 6, "", ("10.0.0.1", 0))]),
+            pytest.raises(OIDCError, match="non-public address.*10.0.0.1"),
+        ):
+            validate_issuer_url("https://internal.corp.example.com")
+
+    def test_rejects_private_172_range(self):
+        """Hostnames resolving to 172.16-31.x.x are rejected."""
+        with (
+            patch("socket.getaddrinfo", return_value=[(2, 1, 6, "", ("172.16.0.1", 0))]),
+            pytest.raises(OIDCError, match="non-public address.*172.16.0.1"),
+        ):
+            validate_issuer_url("https://internal.corp.example.com")
+
+    def test_rejects_private_192_168_range(self):
+        """Hostnames resolving to 192.168.x.x are rejected."""
+        with (
+            patch("socket.getaddrinfo", return_value=[(2, 1, 6, "", ("192.168.1.1", 0))]),
+            pytest.raises(OIDCError, match="non-public address.*192.168.1.1"),
+        ):
+            validate_issuer_url("https://internal.corp.example.com")
+
+    def test_rejects_loopback_127(self):
+        """Hostnames resolving to 127.x.x.x are rejected (non-localhost host)."""
+        with (
+            patch("socket.getaddrinfo", return_value=[(2, 1, 6, "", ("127.0.0.1", 0))]),
+            pytest.raises(OIDCError, match="non-public address.*127.0.0.1"),
+        ):
+            validate_issuer_url("https://evil.example.com")
+
+    def test_rejects_ipv6_loopback(self):
+        """Hostnames resolving to ::1 are rejected (non-localhost host)."""
+        with (
+            patch("socket.getaddrinfo", return_value=[(10, 1, 6, "", ("::1", 0, 0, 0))]),
+            pytest.raises(OIDCError, match="non-public address.*::1"),
+        ):
+            validate_issuer_url("https://evil.example.com")
+
+    def test_rejects_ipv6_private(self):
+        """Hostnames resolving to fc00::/7 are rejected."""
+        with (
+            patch("socket.getaddrinfo", return_value=[(10, 1, 6, "", ("fd00::1", 0, 0, 0))]),
+            pytest.raises(OIDCError, match="non-public address.*fd00::1"),
+        ):
+            validate_issuer_url("https://evil.example.com")
+
+    def test_rejects_link_local(self):
+        """Hostnames resolving to link-local addresses are rejected."""
+        with (
+            patch("socket.getaddrinfo", return_value=[(2, 1, 6, "", ("169.254.169.254", 0))]),
+            pytest.raises(OIDCError, match="non-public address.*169.254.169.254"),
+        ):
+            validate_issuer_url("https://metadata.internal")
+
+    def test_rejects_unresolvable_hostname(self):
+        """DNS resolution failure is rejected."""
+        import socket as _socket
+
+        with (
+            patch("socket.getaddrinfo", side_effect=_socket.gaierror("not found")),
+            pytest.raises(OIDCError, match="cannot be resolved"),
+        ):
+            validate_issuer_url("https://nonexistent.invalid")
+
+    def test_rejects_mixed_addresses(self):
+        """If any resolved address is private, the URL is rejected."""
+        with (
+            patch(
+                "socket.getaddrinfo",
+                return_value=[
+                    (2, 1, 6, "", ("93.184.216.34", 0)),
+                    (2, 1, 6, "", ("10.0.0.1", 0)),
+                ],
+            ),
+            pytest.raises(OIDCError, match="non-public address.*10.0.0.1"),
+        ):
+            validate_issuer_url("https://dual-homed.example.com")
+
+    def test_discover_rejects_ssrf(self):
+        """discover_oidc returns enabled=False when issuer URL fails SSRF check."""
+        config = _make_config(
+            issuer="http://10.0.0.1:8080",
+            authorization_endpoint="",
+            token_endpoint="",
+            userinfo_endpoint="",
+            jwks_uri="",
+        )
+
+        async def _run():
+            result = await discover_oidc(config)
+            assert result.enabled is False
+
+        asyncio.run(_run())
+
+
 # ---------------------------------------------------------------------------
 # Redirect URI Builder
 # ---------------------------------------------------------------------------
@@ -869,6 +1026,9 @@ class TestApplyRoleMapping:
 
 
 class TestDiscoverOIDC:
+    # Mock DNS result for a public IP — reused across discovery tests.
+    _PUBLIC_ADDR = [(2, 1, 6, "", ("93.184.216.34", 0))]
+
     def test_discover_oidc_success(self):
         """Mock httpx response, verify endpoints populated."""
         config = _make_config(
@@ -891,7 +1051,10 @@ class TestDiscoverOIDC:
 
         async def _run():
             client = _mock_async_client(lambda url: _async_return(mock_response))
-            with patch("httpx.AsyncClient", return_value=client):
+            with (
+                patch("socket.getaddrinfo", return_value=self._PUBLIC_ADDR),
+                patch("httpx.AsyncClient", return_value=client),
+            ):
                 result = await discover_oidc(config)
 
             assert result.authorization_endpoint == "https://idp.example.com/authorize"
@@ -916,7 +1079,10 @@ class TestDiscoverOIDC:
 
         async def _run():
             client = _mock_async_client(_failing_get)
-            with patch("httpx.AsyncClient", return_value=client):
+            with (
+                patch("socket.getaddrinfo", return_value=self._PUBLIC_ADDR),
+                patch("httpx.AsyncClient", return_value=client),
+            ):
                 result = await discover_oidc(config)
 
             assert result.enabled is False
@@ -954,7 +1120,10 @@ class TestDiscoverOIDC:
 
         async def _run():
             client = _mock_async_client(lambda url: _async_return(mock_response))
-            with patch("httpx.AsyncClient", return_value=client):
+            with (
+                patch("socket.getaddrinfo", return_value=self._PUBLIC_ADDR),
+                patch("httpx.AsyncClient", return_value=client),
+            ):
                 result = await discover_oidc(config)
 
             assert result.enabled is False

tests/test_oidc_handlers.py

@@ -131,7 +131,7 @@ def authorize_client(storage: SQLiteBackend, oidc_config: OIDCConfig) -> TestCli
     )
     app.state.oidc_config = oidc_config
     app.state.auth_storage = storage
-    app.state.jwt_secret = "test-jwt-secret"
+    app.state.jwt_secret = "test-jwt-secret-key-padded-32b!!"
     app.state.jwks_data = {"keys": []}
     app.state.login_limiter = None
     return TestClient(app, raise_server_exceptions=False)
@@ -468,7 +468,7 @@ class TestOIDCCallback:
         )
         app.state.oidc_config = _make_oidc_config()
         app.state.auth_storage = backend
-        app.state.jwt_secret = "secret"
+        app.state.jwt_secret = "test-jwt-secret-key-padded-32b!!"
         app.state.jwks_data = {"keys": []}
         app.state.login_limiter = None
 
@@ -492,7 +492,7 @@ class TestOIDCCallback:
         )
         app.state.oidc_config = _make_oidc_config()
         app.state.auth_storage = storage
-        app.state.jwt_secret = "secret"
+        app.state.jwt_secret = "test-jwt-secret-key-padded-32b!!"
         app.state.jwks_data = {"keys": []}
         limiter = LoginRateLimiter(max_attempts=1, window_seconds=300)
         limiter.record("ip:testclient")

tests/test_oidc_storage.py

@@ -6,15 +6,6 @@ import time
 
 import pytest
 
-from turnstone.core.storage._sqlite import SQLiteBackend
-
-
-@pytest.fixture()
-def db(tmp_path):
-    """Create a fresh SQLite backend for each test."""
-    return SQLiteBackend(str(tmp_path / "test.db"))
-
-
 # ---------------------------------------------------------------------------
 # OIDC Identity CRUD
 # ---------------------------------------------------------------------------

tests/test_output_assessment_storage.py

@@ -4,16 +4,6 @@ from __future__ import annotations
 
 from datetime import UTC, datetime, timedelta
 
-import pytest
-
-from turnstone.core.storage._sqlite import SQLiteBackend
-
-
-@pytest.fixture()
-def db(tmp_path):
-    """Fresh SQLite backend for each test."""
-    return SQLiteBackend(str(tmp_path / "test.db"))
-
 
 def _make_assessment_kwargs(**overrides):
     """Build default kwargs for record_output_assessment."""

tests/test_scheduled_tasks_storage.py

@@ -4,17 +4,6 @@ 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."""

tests/test_sdk_governance.py

@@ -0,0 +1,440 @@
+"""Integration tests for SDK governance methods against a real Starlette app.
+
+Verifies round-trip serialization: SDK -> HTTP -> Starlette handler -> storage
+-> JSON response -> Pydantic model validation in the SDK client.
+"""
+
+from __future__ import annotations
+
+from typing import TYPE_CHECKING, Any
+
+import httpx
+import pytest
+from starlette.applications import Starlette
+from starlette.middleware import Middleware
+from starlette.middleware.base import BaseHTTPMiddleware
+from starlette.routing import Mount, Route
+
+if TYPE_CHECKING:
+    from starlette.requests import Request
+    from starlette.responses import Response
+
+from turnstone.api.console_schemas import (
+    ListOrgsResponse,
+    ListRolesResponse,
+    ListToolPoliciesResponse,
+    OrgInfo,
+    RoleInfo,
+    ToolPolicyInfo,
+)
+from turnstone.api.schemas import StatusResponse
+from turnstone.console.server import (
+    admin_create_policy,
+    admin_create_role,
+    admin_delete_policy,
+    admin_delete_role,
+    admin_get_org,
+    admin_list_orgs,
+    admin_list_policies,
+    admin_list_roles,
+    admin_update_org,
+    admin_update_policy,
+    admin_update_role,
+)
+from turnstone.core.auth import AuthResult
+from turnstone.core.storage._sqlite import SQLiteBackend
+from turnstone.sdk.console import AsyncTurnstoneConsole
+
+# ---------------------------------------------------------------------------
+# Auth bypass middleware — injects a full-access AuthResult on every request.
+# ---------------------------------------------------------------------------
+
+
+class _InjectAuthMiddleware(BaseHTTPMiddleware):
+    async def dispatch(self, request: Request, call_next: Any) -> Response:
+        request.state.auth_result = AuthResult(
+            user_id="test-admin",
+            scopes=frozenset({"approve"}),
+            token_source="config",
+            permissions=frozenset(
+                {
+                    "read",
+                    "write",
+                    "approve",
+                    "admin.roles",
+                    "admin.orgs",
+                    "admin.policies",
+                }
+            ),
+        )
+        resp: Response = await call_next(request)
+        return resp
+
+
+# ---------------------------------------------------------------------------
+# Fixtures
+# ---------------------------------------------------------------------------
+
+
+def _make_app() -> Starlette:
+    return Starlette(
+        routes=[
+            Mount(
+                "/v1",
+                routes=[
+                    # Roles
+                    Route("/api/admin/roles", admin_list_roles),
+                    Route("/api/admin/roles", admin_create_role, methods=["POST"]),
+                    Route("/api/admin/roles/{role_id}", admin_update_role, methods=["PUT"]),
+                    Route("/api/admin/roles/{role_id}", admin_delete_role, methods=["DELETE"]),
+                    # Orgs
+                    Route("/api/admin/orgs", admin_list_orgs),
+                    Route("/api/admin/orgs/{org_id}", admin_get_org),
+                    Route("/api/admin/orgs/{org_id}", admin_update_org, methods=["PUT"]),
+                    # Policies
+                    Route("/api/admin/policies", admin_list_policies),
+                    Route("/api/admin/policies", admin_create_policy, methods=["POST"]),
+                    Route(
+                        "/api/admin/policies/{policy_id}",
+                        admin_update_policy,
+                        methods=["PUT"],
+                    ),
+                    Route(
+                        "/api/admin/policies/{policy_id}",
+                        admin_delete_policy,
+                        methods=["DELETE"],
+                    ),
+                ],
+            ),
+        ],
+        middleware=[Middleware(_InjectAuthMiddleware)],
+    )
+
+
+@pytest.fixture
+def storage(tmp_path: Any) -> SQLiteBackend:
+    return SQLiteBackend(str(tmp_path / "test.db"))
+
+
+@pytest.fixture
+async def sdk_client(storage: SQLiteBackend):
+    """SDK client wired to a real Starlette app via ASGITransport."""
+    app = _make_app()
+    app.state.auth_storage = storage
+    transport = httpx.ASGITransport(app=app)
+    async with httpx.AsyncClient(transport=transport, base_url="http://testserver") as hc:
+        yield AsyncTurnstoneConsole(httpx_client=hc)
+
+
+# ---------------------------------------------------------------------------
+# Tests — Roles round-trip
+# ---------------------------------------------------------------------------
+
+
+class TestRolesRoundTrip:
+    @pytest.mark.anyio
+    async def test_list_roles_empty(self, sdk_client: AsyncTurnstoneConsole) -> None:
+        resp = await sdk_client.list_roles()
+        assert isinstance(resp, ListRolesResponse)
+        assert resp.roles == []
+
+    @pytest.mark.anyio
+    async def test_create_and_list_role(self, sdk_client: AsyncTurnstoneConsole) -> None:
+        role = await sdk_client.create_role(
+            "analyst", display_name="Data Analyst", permissions="read,write"
+        )
+        assert isinstance(role, RoleInfo)
+        assert role.name == "analyst"
+        assert role.display_name == "Data Analyst"
+        assert role.permissions == "read,write"
+        assert role.builtin is False
+        assert role.role_id  # non-empty
+
+        # List should now contain the new role
+        resp = await sdk_client.list_roles()
+        assert len(resp.roles) == 1
+        assert resp.roles[0].role_id == role.role_id
+        assert resp.roles[0].name == "analyst"
+
+    @pytest.mark.anyio
+    async def test_create_update_role(self, sdk_client: AsyncTurnstoneConsole) -> None:
+        role = await sdk_client.create_role("ops", permissions="read")
+        assert role.permissions == "read"
+
+        updated = await sdk_client.update_role(
+            role.role_id, display_name="Operations", permissions="read,write,approve"
+        )
+        assert isinstance(updated, RoleInfo)
+        assert updated.display_name == "Operations"
+        assert updated.permissions == "read,write,approve"
+        assert updated.role_id == role.role_id
+
+    @pytest.mark.anyio
+    async def test_create_delete_role(self, sdk_client: AsyncTurnstoneConsole) -> None:
+        role = await sdk_client.create_role("temp-role", permissions="read")
+
+        result = await sdk_client.delete_role(role.role_id)
+        assert isinstance(result, StatusResponse)
+        assert result.status == "ok"
+
+        # Verify gone
+        resp = await sdk_client.list_roles()
+        assert resp.roles == []
+
+    @pytest.mark.anyio
+    async def test_full_lifecycle(self, sdk_client: AsyncTurnstoneConsole) -> None:
+        """Create -> list -> update -> list -> delete -> list."""
+        # Create
+        role = await sdk_client.create_role(
+            "lifecycle", display_name="Lifecycle", permissions="read"
+        )
+        role_id = role.role_id
+
+        # List confirms creation
+        roles = (await sdk_client.list_roles()).roles
+        assert len(roles) == 1
+        assert roles[0].role_id == role_id
+
+        # Update
+        updated = await sdk_client.update_role(role_id, permissions="read,write")
+        assert updated.permissions == "read,write"
+
+        # List still has one
+        roles = (await sdk_client.list_roles()).roles
+        assert len(roles) == 1
+        assert roles[0].permissions == "read,write"
+
+        # Delete
+        await sdk_client.delete_role(role_id)
+
+        # List is empty
+        roles = (await sdk_client.list_roles()).roles
+        assert roles == []
+
+    @pytest.mark.anyio
+    async def test_delete_nonexistent_role_raises(self, sdk_client: AsyncTurnstoneConsole) -> None:
+        from turnstone.sdk._types import TurnstoneAPIError
+
+        with pytest.raises(TurnstoneAPIError) as exc_info:
+            await sdk_client.delete_role("nonexistent")
+        assert exc_info.value.status_code == 404
+
+    @pytest.mark.anyio
+    async def test_update_nonexistent_role_raises(self, sdk_client: AsyncTurnstoneConsole) -> None:
+        from turnstone.sdk._types import TurnstoneAPIError
+
+        with pytest.raises(TurnstoneAPIError) as exc_info:
+            await sdk_client.update_role("nonexistent", display_name="Nope")
+        assert exc_info.value.status_code == 404
+
+
+# ---------------------------------------------------------------------------
+# Tests — Policies round-trip
+# ---------------------------------------------------------------------------
+
+
+class TestPoliciesRoundTrip:
+    @pytest.mark.anyio
+    async def test_list_policies_empty(self, sdk_client: AsyncTurnstoneConsole) -> None:
+        resp = await sdk_client.list_policies()
+        assert isinstance(resp, ListToolPoliciesResponse)
+        assert resp.policies == []
+
+    @pytest.mark.anyio
+    async def test_create_and_list_policy(self, sdk_client: AsyncTurnstoneConsole) -> None:
+        policy = await sdk_client.create_policy("Allow bash", "bash_*", "allow", priority=10)
+        assert isinstance(policy, ToolPolicyInfo)
+        assert policy.name == "Allow bash"
+        assert policy.tool_pattern == "bash_*"
+        assert policy.action == "allow"
+        assert policy.priority == 10
+        assert policy.enabled is True
+        assert policy.policy_id  # non-empty
+
+        resp = await sdk_client.list_policies()
+        assert len(resp.policies) == 1
+        assert resp.policies[0].policy_id == policy.policy_id
+
+    @pytest.mark.anyio
+    async def test_create_update_policy(self, sdk_client: AsyncTurnstoneConsole) -> None:
+        policy = await sdk_client.create_policy("Deny write", "write_*", "deny", priority=5)
+
+        updated = await sdk_client.update_policy(
+            policy.policy_id, name="Allow write", action="allow", priority=20
+        )
+        assert isinstance(updated, ToolPolicyInfo)
+        assert updated.name == "Allow write"
+        assert updated.action == "allow"
+        assert updated.priority == 20
+        assert updated.policy_id == policy.policy_id
+
+    @pytest.mark.anyio
+    async def test_create_delete_policy(self, sdk_client: AsyncTurnstoneConsole) -> None:
+        policy = await sdk_client.create_policy("Temp policy", "temp_*", "ask")
+
+        result = await sdk_client.delete_policy(policy.policy_id)
+        assert isinstance(result, StatusResponse)
+        assert result.status == "ok"
+
+        resp = await sdk_client.list_policies()
+        assert resp.policies == []
+
+    @pytest.mark.anyio
+    async def test_full_lifecycle(self, sdk_client: AsyncTurnstoneConsole) -> None:
+        """Create -> list -> update -> list -> delete -> list."""
+        policy = await sdk_client.create_policy("Lifecycle", "test_*", "deny", priority=1)
+        pid = policy.policy_id
+
+        policies = (await sdk_client.list_policies()).policies
+        assert len(policies) == 1
+
+        await sdk_client.update_policy(pid, action="allow", priority=99)
+        policies = (await sdk_client.list_policies()).policies
+        assert policies[0].action == "allow"
+        assert policies[0].priority == 99
+
+        await sdk_client.delete_policy(pid)
+        policies = (await sdk_client.list_policies()).policies
+        assert policies == []
+
+    @pytest.mark.anyio
+    async def test_delete_nonexistent_policy_raises(
+        self, sdk_client: AsyncTurnstoneConsole
+    ) -> None:
+        from turnstone.sdk._types import TurnstoneAPIError
+
+        with pytest.raises(TurnstoneAPIError) as exc_info:
+            await sdk_client.delete_policy("nonexistent")
+        assert exc_info.value.status_code == 404
+
+    @pytest.mark.anyio
+    async def test_update_nonexistent_policy_raises(
+        self, sdk_client: AsyncTurnstoneConsole
+    ) -> None:
+        from turnstone.sdk._types import TurnstoneAPIError
+
+        with pytest.raises(TurnstoneAPIError) as exc_info:
+            await sdk_client.update_policy("nonexistent", name="Nope")
+        assert exc_info.value.status_code == 404
+
+    @pytest.mark.anyio
+    async def test_create_policy_invalid_action_raises(
+        self, sdk_client: AsyncTurnstoneConsole
+    ) -> None:
+        from turnstone.sdk._types import TurnstoneAPIError
+
+        with pytest.raises(TurnstoneAPIError) as exc_info:
+            await sdk_client.create_policy("Bad", "tool_*", "yolo")
+        assert exc_info.value.status_code == 400
+
+
+# ---------------------------------------------------------------------------
+# Tests — Orgs round-trip
+# ---------------------------------------------------------------------------
+
+
+class TestOrgsRoundTrip:
+    @pytest.mark.anyio
+    async def test_list_orgs_empty(self, sdk_client: AsyncTurnstoneConsole) -> None:
+        resp = await sdk_client.list_orgs()
+        assert isinstance(resp, ListOrgsResponse)
+        assert resp.orgs == []
+
+    @pytest.mark.anyio
+    async def test_get_org(self, sdk_client: AsyncTurnstoneConsole, storage: SQLiteBackend) -> None:
+        storage.create_org(
+            org_id="org-1", name="acme", display_name="Acme Corp", settings='{"k": "v"}'
+        )
+        org = await sdk_client.get_org("org-1")
+        assert isinstance(org, OrgInfo)
+        assert org.org_id == "org-1"
+        assert org.name == "acme"
+        assert org.display_name == "Acme Corp"
+        assert org.settings == '{"k": "v"}'
+
+    @pytest.mark.anyio
+    async def test_list_orgs_after_seed(
+        self, sdk_client: AsyncTurnstoneConsole, storage: SQLiteBackend
+    ) -> None:
+        storage.create_org(org_id="org-a", name="alpha", display_name="Alpha")
+        storage.create_org(org_id="org-b", name="beta", display_name="Beta")
+
+        resp = await sdk_client.list_orgs()
+        assert len(resp.orgs) == 2
+        names = {o.name for o in resp.orgs}
+        assert names == {"alpha", "beta"}
+
+    @pytest.mark.anyio
+    async def test_update_org(
+        self, sdk_client: AsyncTurnstoneConsole, storage: SQLiteBackend
+    ) -> None:
+        storage.create_org(org_id="org-1", name="acme", display_name="Acme Corp")
+
+        updated = await sdk_client.update_org("org-1", display_name="Acme Inc.")
+        assert isinstance(updated, OrgInfo)
+        assert updated.display_name == "Acme Inc."
+        assert updated.org_id == "org-1"
+
+    @pytest.mark.anyio
+    async def test_get_nonexistent_org_raises(self, sdk_client: AsyncTurnstoneConsole) -> None:
+        from turnstone.sdk._types import TurnstoneAPIError
+
+        with pytest.raises(TurnstoneAPIError) as exc_info:
+            await sdk_client.get_org("nonexistent")
+        assert exc_info.value.status_code == 404
+
+    @pytest.mark.anyio
+    async def test_update_nonexistent_org_raises(self, sdk_client: AsyncTurnstoneConsole) -> None:
+        from turnstone.sdk._types import TurnstoneAPIError
+
+        with pytest.raises(TurnstoneAPIError) as exc_info:
+            await sdk_client.update_org("nonexistent", display_name="Nope")
+        assert exc_info.value.status_code == 404
+
+
+# ---------------------------------------------------------------------------
+# Tests — Pydantic model field validation
+# ---------------------------------------------------------------------------
+
+
+class TestModelValidation:
+    """Verify that all expected fields are populated and correctly typed."""
+
+    @pytest.mark.anyio
+    async def test_role_info_fields(self, sdk_client: AsyncTurnstoneConsole) -> None:
+        role = await sdk_client.create_role("reviewer", permissions="read")
+        assert isinstance(role.role_id, str)
+        assert isinstance(role.name, str)
+        assert isinstance(role.display_name, str)
+        assert isinstance(role.permissions, str)
+        assert isinstance(role.builtin, bool)
+        assert isinstance(role.org_id, str)
+        assert isinstance(role.created, str)
+        assert isinstance(role.updated, str)
+
+    @pytest.mark.anyio
+    async def test_policy_info_fields(self, sdk_client: AsyncTurnstoneConsole) -> None:
+        policy = await sdk_client.create_policy("Test", "read_*", "allow", priority=5)
+        assert isinstance(policy.policy_id, str)
+        assert isinstance(policy.name, str)
+        assert isinstance(policy.tool_pattern, str)
+        assert isinstance(policy.action, str)
+        assert isinstance(policy.priority, int)
+        assert isinstance(policy.org_id, str)
+        assert isinstance(policy.enabled, bool)
+        assert isinstance(policy.created_by, str)
+        assert isinstance(policy.created, str)
+        assert isinstance(policy.updated, str)
+
+    @pytest.mark.anyio
+    async def test_org_info_fields(
+        self, sdk_client: AsyncTurnstoneConsole, storage: SQLiteBackend
+    ) -> None:
+        storage.create_org(org_id="org-v", name="validate", display_name="Validate")
+        org = await sdk_client.get_org("org-v")
+        assert isinstance(org.org_id, str)
+        assert isinstance(org.name, str)
+        assert isinstance(org.display_name, str)
+        assert isinstance(org.settings, str)
+        assert isinstance(org.created, str)
+        assert isinstance(org.updated, str)

tests/test_server_live.py

@@ -623,6 +623,7 @@ class TestServerHealthMetrics:
 
         mock_mgr = MagicMock()
         mock_mgr.list_all.return_value = [mock_ws]
+        mock_mgr.max_workstreams = 10
 
         app = srv_mod.create_app(
             workstreams=mock_mgr,
@@ -799,6 +800,7 @@ class TestServerRateLimiting:
 
         mock_mgr = MagicMock()
         mock_mgr.list_all.return_value = [mock_ws]
+        mock_mgr.max_workstreams = 10
 
         app = srv_mod.create_app(
             workstreams=mock_mgr,

tests/test_services_storage.py

@@ -2,15 +2,6 @@
 
 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):

tests/test_session.py

@@ -214,7 +214,7 @@ class TestPlanExec:
                         "id": tc_id,
                         "type": "function",
                         "function": {
-                            "name": "create_plan",
+                            "name": "plan_agent",
                             "arguments": json.dumps({"goal": prior_prompt}),
                         },
                     }
@@ -250,7 +250,7 @@ class TestPlanExec:
             m for m in messages if m["role"] == "assistant" and m.get("tool_calls")
         ]
         assert len(assistant_with_tc) == 1
-        assert assistant_with_tc[0]["tool_calls"][0]["function"]["name"] == "create_plan"
+        assert assistant_with_tc[0]["tool_calls"][0]["function"]["name"] == "plan_agent"
 
         # The real tool result is forwarded with its original content
         tool_msgs = [m for m in messages if m["role"] == "tool"]
@@ -463,7 +463,7 @@ class TestPlanRefinement:
         with patch.object(session, "_refine_plan", side_effect=fake_refine):
             items = [
                 {
-                    "func_name": "create_plan",
+                    "func_name": "plan_agent",
                     "call_id": "c1",
                     "prompt": "add auth",
                 }
@@ -576,7 +576,7 @@ class TestPlanRefinement:
         msgs = captured["messages"]
         assert msgs[0]["role"] == "system"
         assert msgs[1]["role"] == "assistant"
-        assert msgs[1]["tool_calls"][0]["function"]["name"] == "create_plan"
+        assert msgs[1]["tool_calls"][0]["function"]["name"] == "plan_agent"
         assert msgs[2]["role"] == "tool"
         assert msgs[2]["content"] == self.GOOD_PLAN
         assert msgs[3]["role"] == "user"
@@ -753,3 +753,268 @@ class TestGetCapabilitiesOverride:
         caps = session._get_capabilities()
         # Default OpenAI provider for unknown model → no vision
         assert caps.supports_vision is False
+
+
+class TestTitleRetry:
+    """_generate_title resets _title_generated on failure."""
+
+    def test_title_generated_reset_on_failure(self, tmp_db):
+        session = _make_session()
+        session._title_generated = True
+        session.messages = [
+            {"role": "user", "content": "Hello"},
+            {"role": "assistant", "content": "Hi there"},
+        ]
+        # Mock provider to raise
+        session._provider = MagicMock()
+        session._provider.create_completion.side_effect = RuntimeError("API error")
+
+        session._generate_title()
+
+        assert session._title_generated is False
+
+    def test_title_generated_stays_true_on_success(self, tmp_db):
+        session = _make_session()
+        session._title_generated = True
+        session.messages = [
+            {"role": "user", "content": "Hello"},
+            {"role": "assistant", "content": "Hi there"},
+        ]
+        result = MagicMock()
+        result.content = "Test Title"
+        session._provider = MagicMock()
+        session._provider.create_completion.return_value = result
+
+        with patch("turnstone.core.session.update_workstream_title"):
+            session._generate_title()
+
+        # Flag stays True after successful generation
+        assert session._title_generated is True
+
+    def test_title_skipped_after_resume_changes_ws_id(self, tmp_db):
+        """If ws_id changes (via resume) during title generation, discard the result."""
+        session = _make_session()
+        session._title_generated = True
+        session.messages = [
+            {"role": "user", "content": "Hello"},
+            {"role": "assistant", "content": "Hi there"},
+        ]
+        original_ws_id = session._ws_id
+        result = MagicMock()
+        result.content = "Test Title"
+        session._provider = MagicMock()
+        session._provider.create_completion.return_value = result
+
+        # Simulate resume() changing ws_id while title generation is in flight
+        def _change_ws_id(*args, **kwargs):
+            session._ws_id = "different-ws-id"
+            return result
+
+        session._provider.create_completion.side_effect = _change_ws_id
+
+        with patch("turnstone.core.session.update_workstream_title") as mock_update:
+            session._generate_title()
+
+        # Title should NOT be applied to the new workstream
+        mock_update.assert_not_called()
+        # Restore for cleanup
+        session._ws_id = original_ws_id
+
+
+class TestLiveConfigUpdate:
+    """ConfigStore-backed sessions pick up settings changes at point-of-use."""
+
+    def test_memory_config_reads_from_config_store(self, tmp_db):
+        """_mem_cfg returns live values from ConfigStore when present."""
+        from turnstone.core.config_store import ConfigStore
+        from turnstone.core.storage._sqlite import SQLiteBackend
+
+        storage = SQLiteBackend(str(tmp_db), create_tables=True)
+        cs = ConfigStore(storage)
+        session = _make_session(config_store=cs)
+
+        # Default: relevance_k=5
+        assert session._mem_cfg.relevance_k == 5
+
+        # Admin changes the setting
+        cs.set("memory.relevance_k", 10, changed_by="test")
+        assert session._mem_cfg.relevance_k == 10
+
+    def test_judge_config_reads_from_config_store(self, tmp_db):
+        """_judge_cfg returns live behavioral flags from ConfigStore."""
+        from turnstone.core.config_store import ConfigStore
+        from turnstone.core.judge import JudgeConfig
+        from turnstone.core.storage._sqlite import SQLiteBackend
+
+        storage = SQLiteBackend(str(tmp_db), create_tables=True)
+        cs = ConfigStore(storage)
+        session = _make_session(
+            judge_config=JudgeConfig(),
+            config_store=cs,
+        )
+
+        # Default: enabled=True
+        assert session._judge_cfg.enabled is True
+
+        # Admin disables the judge
+        cs.set("judge.enabled", False, changed_by="test")
+        assert session._judge_cfg.enabled is False
+
+    def test_judge_client_config_stays_frozen(self, tmp_db):
+        """LLM client fields (model, provider) are frozen from creation time."""
+        from turnstone.core.config_store import ConfigStore
+        from turnstone.core.judge import JudgeConfig
+        from turnstone.core.storage._sqlite import SQLiteBackend
+
+        storage = SQLiteBackend(str(tmp_db), create_tables=True)
+        cs = ConfigStore(storage)
+        session = _make_session(
+            judge_config=JudgeConfig(model="original-model"),
+            config_store=cs,
+        )
+
+        # Change the model in ConfigStore — should NOT affect the session
+        cs.set("judge.model", "new-model", changed_by="test")
+        assert session._judge_cfg.model == "original-model"
+
+    def test_judge_disable_after_init_stops_future_use(self, tmp_db):
+        """Disabling judge.enabled after IntentJudge is created returns None."""
+        from turnstone.core.config_store import ConfigStore
+        from turnstone.core.judge import JudgeConfig
+        from turnstone.core.storage._sqlite import SQLiteBackend
+
+        storage = SQLiteBackend(str(tmp_db), create_tables=True)
+        cs = ConfigStore(storage)
+        session = _make_session(
+            judge_config=JudgeConfig(),
+            config_store=cs,
+        )
+
+        # Force judge initialization by setting a mock
+        session._judge = MagicMock()
+        assert session._ensure_judge() is not None
+
+        # Admin disables the judge — cached instance should NOT be returned
+        cs.set("judge.enabled", False, changed_by="test")
+        assert session._ensure_judge() is None
+
+    def test_fallback_to_frozen_without_config_store(self, tmp_db):
+        """Without ConfigStore (CLI mode), frozen config is used."""
+        from turnstone.core.memory_relevance import MemoryConfig
+
+        session = _make_session(memory_config=MemoryConfig(relevance_k=3))
+        assert session._mem_cfg.relevance_k == 3
+
+
+class TestAgentOutputGuard:
+    """Output guard should evaluate tool results in _run_agent, not just the main loop."""
+
+    def test_agent_loop_calls_evaluate_output(self):
+        """_run_agent passes tool output through _evaluate_output when output_guard is enabled."""
+        from turnstone.core.judge import JudgeConfig
+
+        session = _make_session(judge_config=JudgeConfig(output_guard=True))
+
+        with patch.object(session, "_evaluate_output", wraps=lambda cid, o, fn: o) as mock_eval:
+            # Simulate _run_agent getting a tool call response then a text response
+            call_count = [0]
+
+            def fake_create(**kwargs):
+                call_count[0] += 1
+                resp = MagicMock()
+                if call_count[0] == 1:
+                    # First call: model returns a tool call
+                    choice = MagicMock()
+                    choice.finish_reason = "tool_calls"
+                    tc = MagicMock()
+                    tc.id = "call_1"
+                    tc.function.name = "read_file"
+                    tc.function.arguments = '{"path": "/tmp/test"}'
+                    choice.message.tool_calls = [tc]
+                    choice.message.content = None
+                    resp.choices = [choice]
+                    resp.usage = MagicMock(prompt_tokens=10, completion_tokens=5)
+                else:
+                    # Second call: model returns text (done)
+                    choice = MagicMock()
+                    choice.finish_reason = "stop"
+                    choice.message.tool_calls = None
+                    choice.message.content = "Done"
+                    resp.choices = [choice]
+                    resp.usage = MagicMock(prompt_tokens=10, completion_tokens=5)
+                return resp
+
+            session.client.chat.completions.create = fake_create
+
+            # Mock tool preparation to return a simple output
+            def fake_prepare(tc_dict, **kwargs):
+                return {
+                    "call_id": tc_dict["id"],
+                    "func_name": "read_file",
+                    "needs_approval": False,
+                    "execute": lambda p: ("call_1", "file contents with sk-proj-SECRET123"),
+                }
+
+            with patch.object(session, "_prepare_tool", side_effect=fake_prepare):
+                session._run_agent(
+                    [{"role": "user", "content": "test"}],
+                    tools=[{"type": "function", "function": {"name": "read_file"}}],
+                    label="test",
+                )
+
+            mock_eval.assert_called_once()
+            args = mock_eval.call_args[0]
+            assert args[0] == "call_1"  # call_id
+            assert "sk-proj-SECRET123" in args[1]  # output
+            assert args[2] == "read_file"  # func_name
+
+    def test_agent_loop_skips_guard_when_disabled(self):
+        """_run_agent does not call _evaluate_output when output_guard is disabled."""
+        from turnstone.core.judge import JudgeConfig
+
+        session = _make_session(judge_config=JudgeConfig(output_guard=False))
+
+        with patch.object(session, "_evaluate_output") as mock_eval:
+            call_count = [0]
+
+            def fake_create(**kwargs):
+                call_count[0] += 1
+                resp = MagicMock()
+                if call_count[0] == 1:
+                    choice = MagicMock()
+                    choice.finish_reason = "tool_calls"
+                    tc = MagicMock()
+                    tc.id = "call_1"
+                    tc.function.name = "read_file"
+                    tc.function.arguments = '{"path": "/tmp/test"}'
+                    choice.message.tool_calls = [tc]
+                    choice.message.content = None
+                    resp.choices = [choice]
+                    resp.usage = MagicMock(prompt_tokens=10, completion_tokens=5)
+                else:
+                    choice = MagicMock()
+                    choice.finish_reason = "stop"
+                    choice.message.tool_calls = None
+                    choice.message.content = "Done"
+                    resp.choices = [choice]
+                    resp.usage = MagicMock(prompt_tokens=10, completion_tokens=5)
+                return resp
+
+            session.client.chat.completions.create = fake_create
+
+            def fake_prepare(tc_dict, **kwargs):
+                return {
+                    "call_id": tc_dict["id"],
+                    "func_name": "read_file",
+                    "needs_approval": False,
+                    "execute": lambda p: ("call_1", "safe output"),
+                }
+
+            with patch.object(session, "_prepare_tool", side_effect=fake_prepare):
+                session._run_agent(
+                    [{"role": "user", "content": "test"}],
+                    tools=[{"type": "function", "function": {"name": "read_file"}}],
+                    label="test",
+                )
+
+            mock_eval.assert_not_called()

tests/test_sessions.py

@@ -607,3 +607,150 @@ class TestPruneWorkstreams:
         # Config rows should be cleaned up
         assert load_workstream_config("orphan_cfg") == {}
         assert load_workstream_config("stale_cfg") == {}
+
+
+# ── Parallel tool exception isolation ────────────────────────────────
+
+
+class TestParallelToolExceptionIsolation:
+    """Bug #117: one tool raising should not kill the entire batch."""
+
+    def test_exception_in_one_tool_does_not_kill_batch(self, tmp_db, mock_openai_client):
+        from unittest.mock import patch
+
+        session = ChatSession(
+            client=mock_openai_client,
+            model="test-model",
+            ui=MagicMock(),
+            instructions=None,
+            temperature=0.5,
+            max_tokens=1000,
+            tool_timeout=10,
+        )
+
+        def succeed(item):
+            return item["call_id"], "ok"
+
+        def fail(item):
+            raise RuntimeError("boom")
+
+        items = [
+            {
+                "call_id": "c1",
+                "func_name": "bash",
+                "execute": succeed,
+                "needs_approval": False,
+                "header": "test",
+                "preview": "",
+            },
+            {
+                "call_id": "c2",
+                "func_name": "math",
+                "execute": fail,
+                "needs_approval": False,
+                "header": "test",
+                "preview": "",
+            },
+        ]
+
+        tool_calls = [
+            {"id": "c1", "function": {"name": "bash", "arguments": "{}"}},
+            {"id": "c2", "function": {"name": "math", "arguments": "{}"}},
+        ]
+
+        with (
+            patch.object(session, "_prepare_tool", side_effect=items),
+            patch.object(session, "_evaluate_intent"),
+            patch.object(session, "_emit_state"),
+            patch.object(session, "_init_system_messages"),
+            patch.object(session, "_check_cancelled"),
+        ):
+            session.ui.approve_tools.return_value = (True, None)
+            results, _ = session._execute_tools(tool_calls)
+
+        assert results[0] == ("c1", "ok")
+        assert results[1][0] == "c2"
+        assert "Error executing math" in results[1][1]
+        assert "boom" in results[1][1]
+
+
+# ── Web search tool gating ───────────────────────────────────────────
+
+
+class TestWebSearchGating:
+    """Bug #117: web_search should not be offered without a backend."""
+
+    def test_web_search_filtered_when_no_backend(self, tmp_db, mock_openai_client):
+        from unittest.mock import patch
+
+        from turnstone.core.providers._protocol import ModelCapabilities
+
+        session = ChatSession(
+            client=mock_openai_client,
+            model="local-model",
+            ui=MagicMock(),
+            instructions=None,
+            temperature=0.5,
+            max_tokens=1000,
+            tool_timeout=10,
+        )
+
+        caps = ModelCapabilities(supports_web_search=False)
+        with (
+            patch.object(session, "_get_capabilities", return_value=caps),
+            patch("turnstone.core.session.get_tavily_key", return_value=None),
+        ):
+            tools = session._get_active_tools()
+
+        names = [t.get("function", {}).get("name") for t in tools]
+        assert "web_search" not in names
+
+    def test_web_search_kept_when_tavily_available(self, tmp_db, mock_openai_client):
+        from unittest.mock import patch
+
+        from turnstone.core.providers._protocol import ModelCapabilities
+
+        session = ChatSession(
+            client=mock_openai_client,
+            model="local-model",
+            ui=MagicMock(),
+            instructions=None,
+            temperature=0.5,
+            max_tokens=1000,
+            tool_timeout=10,
+        )
+
+        caps = ModelCapabilities(supports_web_search=False)
+        with (
+            patch.object(session, "_get_capabilities", return_value=caps),
+            patch("turnstone.core.session.get_tavily_key", return_value="tvly-test-key"),
+        ):
+            tools = session._get_active_tools()
+
+        names = [t.get("function", {}).get("name") for t in tools]
+        assert "web_search" in names
+
+    def test_web_search_kept_when_native_support(self, tmp_db, mock_openai_client):
+        from unittest.mock import patch
+
+        from turnstone.core.providers._protocol import ModelCapabilities
+
+        session = ChatSession(
+            client=mock_openai_client,
+            model="gpt-5-search-api",
+            ui=MagicMock(),
+            instructions=None,
+            temperature=0.5,
+            max_tokens=1000,
+            tool_timeout=10,
+        )
+
+        caps = ModelCapabilities(supports_web_search=True)
+        with (
+            patch.object(session, "_get_capabilities", return_value=caps),
+            patch("turnstone.core.session.get_tavily_key", return_value=None),
+        ):
+            tools = session._get_active_tools()
+
+        names = [t.get("function", {}).get("name") for t in tools]
+        assert "web_search" in names

tests/test_settings_api.py

@@ -179,7 +179,10 @@ class TestDeleteSetting:
         # Delete it
         r = client.delete("/v1/api/admin/settings/tools.timeout")
         assert r.status_code == 200
-        assert r.json()["status"] == "ok"
+        body = r.json()
+        assert body["status"] == "ok"
+        assert body["key"] == "tools.timeout"
+        assert body["default"] == 120  # registry default for tools.timeout
 
     def test_delete_then_list_shows_default(self, client):
         client.put(

tests/test_sim.py

@@ -74,7 +74,7 @@ class TestSimEngine:
 
     def test_llm_response_returns_content(self, engine):
         async def _test():
-            content, tool_calls = await engine.simulate_llm_response(True, 1)
+            content, tool_calls = await engine.simulate_llm_response(True)
             assert isinstance(content, str)
             assert len(content) > 0
             assert isinstance(tool_calls, list)
@@ -85,8 +85,8 @@ class TestSimEngine:
         async def _test():
             e1 = SimEngine(fast_config, rng=random.Random(123))
             e2 = SimEngine(fast_config, rng=random.Random(123))
-            c1, t1 = await e1.simulate_llm_response(True, 1)
-            c2, t2 = await e2.simulate_llm_response(True, 1)
+            c1, t1 = await e1.simulate_llm_response(True)
+            c2, t2 = await e2.simulate_llm_response(True)
             assert c1 == c2
             assert len(t1) == len(t2)
 

tests/test_skill_parser.py

@@ -18,7 +18,7 @@ description: Automated code review skill
 author: Test Author
 version: 2.0.0
 tags: [python, review, quality]
-allowed_tools: [read_file, list_directory]
+allowed-tools: [read_file, list_directory]
 license: MIT
 compatibility: ">=0.7"
 ---
@@ -187,13 +187,13 @@ Content.
 
 
 class TestAllowedTools:
-    """Verify allowed_tools parsing."""
+    """Verify allowed-tools parsing (Agent Skills standard hyphenated field)."""
 
     def test_list_format(self) -> None:
         raw = """\
 ---
 name: tools-list
-allowed_tools: [bash, read_file]
+allowed-tools: [bash, read_file]
 ---
 
 Content.
@@ -201,11 +201,12 @@ Content.
         result = parse_skill_md(raw)
         assert result.allowed_tools == ["bash", "read_file"]
 
-    def test_csv_format(self) -> None:
+    def test_space_delimited_format(self) -> None:
+        """Standard format per Agent Skills spec."""
         raw = """\
 ---
-name: tools-csv
-allowed_tools: "bash, read_file, write_file"
+name: tools-space
+allowed-tools: "bash read_file write_file"
 ---
 
 Content.
@@ -219,6 +220,19 @@ Content.
 name: no-tools
 ---
 
+Content.
+"""
+        result = parse_skill_md(raw)
+        assert result.allowed_tools == []
+
+    def test_underscore_key_not_read(self) -> None:
+        """allowed_tools (underscore) is not a SKILL.md field — ignored by parser."""
+        raw = """\
+---
+name: legacy-key
+allowed_tools: [bash, read_file]
+---
+
 Content.
 """
         result = parse_skill_md(raw)
@@ -247,3 +261,299 @@ class TestValidateSkillName:
         assert validate_skill_name("HAS-UPPER") is not None
         assert validate_skill_name("has space") is not None
         assert validate_skill_name("-leading-hyphen") is not None
+
+    def test_consecutive_hyphens_rejected(self) -> None:
+        """Agent Skills spec: consecutive hyphens not allowed."""
+        assert validate_skill_name("foo--bar") is not None
+        assert "consecutive hyphens" in (validate_skill_name("a--b") or "")
+        # Single hyphens are fine
+        assert validate_skill_name("foo-bar") is None
+
+
+# -- Agent Skills Standard Compliance Tests -----------------------------------
+
+
+class TestStandardAllowedTools:
+    """Agent Skills spec: 'allowed-tools' (hyphenated), space-delimited."""
+
+    def test_list_format(self) -> None:
+        raw = """\
+---
+name: standard-tools
+allowed-tools: ["Bash(git:*)", "Bash(jq:*)", "Read"]
+---
+
+Content.
+"""
+        result = parse_skill_md(raw)
+        assert result.allowed_tools == ["Bash(git:*)", "Bash(jq:*)", "Read"]
+
+    def test_space_delimited(self) -> None:
+        """Standard format: space-delimited string."""
+        raw = """\
+---
+name: space-tools
+allowed-tools: "Bash(git:*) Bash(jq:*) Read"
+---
+
+Content.
+"""
+        result = parse_skill_md(raw)
+        assert result.allowed_tools == ["Bash(git:*)", "Bash(jq:*)", "Read"]
+
+    def test_mixed_space_comma_delimiters(self) -> None:
+        raw = """\
+---
+name: mixed-delim
+allowed-tools: "Read, Write Bash"
+---
+
+Content.
+"""
+        result = parse_skill_md(raw)
+        assert result.allowed_tools == ["Read", "Write", "Bash"]
+
+
+class TestStandardMetadataNesting:
+    """Standard puts author/version under metadata map."""
+
+    def test_metadata_author(self) -> None:
+        raw = """\
+---
+name: nested-author
+description: Test skill
+metadata:
+  author: example-org
+  version: "2.0"
+---
+
+Content.
+"""
+        result = parse_skill_md(raw)
+        assert result.author == "example-org"
+        assert result.version == "2.0"
+
+    def test_top_level_takes_precedence(self) -> None:
+        raw = """\
+---
+name: precedence
+description: Test skill
+author: top-level
+version: 1.0.0
+metadata:
+  author: nested
+  version: "2.0"
+---
+
+Content.
+"""
+        result = parse_skill_md(raw)
+        assert result.author == "top-level"
+        assert result.version == "1.0.0"
+
+    def test_metadata_version_only(self) -> None:
+        raw = """\
+---
+name: version-only
+description: Test
+metadata:
+  version: "3.5.1"
+---
+
+Content.
+"""
+        result = parse_skill_md(raw)
+        assert result.version == "3.5.1"
+        assert result.author == ""
+
+    def test_null_author_uses_default(self) -> None:
+        """YAML null/bare key must not produce the string 'None'."""
+        raw = """\
+---
+name: null-author
+description: Test
+author:
+---
+
+Content.
+"""
+        result = parse_skill_md(raw)
+        assert result.author == ""
+        assert result.version == "1.0.0"
+
+    def test_null_version_uses_default(self) -> None:
+        raw = """\
+---
+name: null-version
+description: Test
+version:
+---
+
+Content.
+"""
+        result = parse_skill_md(raw)
+        assert result.version == "1.0.0"
+
+    def test_null_description_falls_back_to_body(self) -> None:
+        """YAML null description must not produce 'None' string."""
+        raw = """\
+---
+name: null-desc
+description:
+---
+
+First paragraph here.
+"""
+        result = parse_skill_md(raw)
+        assert result.description == "First paragraph here."
+        assert "None" not in result.description
+
+    def test_null_license_and_compatibility(self) -> None:
+        """YAML null license/compatibility must not produce 'None' string."""
+        raw = """\
+---
+name: null-fields
+description: Test
+license:
+compatibility:
+---
+
+Content.
+"""
+        result = parse_skill_md(raw)
+        assert result.license == ""
+        assert result.compatibility == ""
+
+
+class TestStandardFieldLengths:
+    """Spec caps: description <= 1024, compatibility <= 500."""
+
+    def test_description_truncated_at_1024(self) -> None:
+        long_desc = "x" * 1200
+        raw = f"""\
+---
+name: long-desc
+description: "{long_desc}"
+---
+
+Content.
+"""
+        result = parse_skill_md(raw)
+        assert len(result.description) == 1024
+
+    def test_compatibility_truncated_at_500(self) -> None:
+        long_compat = "y" * 600
+        raw = f"""\
+---
+name: long-compat
+description: Short
+compatibility: "{long_compat}"
+---
+
+Content.
+"""
+        result = parse_skill_md(raw)
+        assert len(result.compatibility) == 500
+
+    def test_short_fields_unchanged(self) -> None:
+        raw = """\
+---
+name: short
+description: Brief
+compatibility: Requires git
+---
+
+Content.
+"""
+        result = parse_skill_md(raw)
+        assert result.description == "Brief"
+        assert result.compatibility == "Requires git"
+
+
+class TestLenientMode:
+    """Lenient parsing for cross-client skill ingestion."""
+
+    def test_invalid_name_sanitized(self) -> None:
+        raw = """\
+---
+name: Invalid_Name!
+description: A test skill
+---
+
+Content.
+"""
+        result = parse_skill_md(raw, lenient=True)
+        assert result is not None
+        assert result.name == "invalidname"
+
+    def test_unsalvageable_name_returns_none(self) -> None:
+        raw = """\
+---
+name: "!!!"
+description: A test skill
+---
+
+Content.
+"""
+        assert parse_skill_md(raw, lenient=True) is None
+
+    def test_missing_description_returns_none(self) -> None:
+        raw = """\
+---
+name: no-desc
+---
+"""
+        assert parse_skill_md(raw, lenient=True) is None
+
+    def test_broken_yaml_returns_none(self) -> None:
+        raw = """\
+---
+name: [broken: yaml: {{{
+---
+
+Content.
+"""
+        assert parse_skill_md(raw, lenient=True) is None
+
+    def test_malformed_yaml_colon_in_description_recovers(self) -> None:
+        """Standard recommends retrying unquoted colon values."""
+        raw = """\
+---
+name: colon-desc
+description: Use this skill when: the user asks about PDFs
+---
+
+Content.
+"""
+        result = parse_skill_md(raw, lenient=True)
+        # The frontmatter library may parse this fine, but if not,
+        # the retry mechanism should recover.
+        assert result is not None
+        assert result.name == "colon-desc"
+        assert "PDF" in result.description
+
+    def test_strict_mode_still_raises(self) -> None:
+        """Default strict mode unchanged."""
+        raw = """\
+---
+name: Invalid_Name!
+description: A test skill
+---
+
+Content.
+"""
+        with pytest.raises(ValueError):
+            parse_skill_md(raw)
+
+    def test_consecutive_hyphens_lenient(self) -> None:
+        raw = """\
+---
+name: foo--bar
+description: A test skill
+---
+
+Content.
+"""
+        result = parse_skill_md(raw, lenient=True)
+        assert result is not None
+        assert "--" not in result.name

tests/test_skill_resources_storage.py

@@ -4,16 +4,6 @@ from __future__ import annotations
 
 import uuid
 
-import pytest
-
-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"))
-
 
 class TestDeleteSkillResourceByPath:
     def test_delete_existing(self, storage):

tests/test_skills.py

@@ -132,6 +132,7 @@ def _create_template(db, template_id, name, content, **kwargs):
         notify_on_complete=kwargs.get("notify_on_complete", "{}"),
         enabled=kwargs.get("enabled", True),
         allowed_tools=kwargs.get("allowed_tools", "[]"),
+        priority=kwargs.get("priority", 0),
     )
 
 
@@ -293,7 +294,7 @@ class TestSkillStorage:
         assert result == []
 
     def test_list_skills_by_activation_ordered_by_name(self, db):
-        """Results are ordered by name ascending."""
+        """Results are ordered by name ascending when priority is equal."""
         _create_template(db, "s2", "beta-search", "B", activation="search")
         _create_template(db, "s1", "alpha-search", "A", activation="search")
         results = db.list_skills_by_activation("search")
@@ -301,6 +302,51 @@ class TestSkillStorage:
         assert results[0]["name"] == "alpha-search"
         assert results[1]["name"] == "beta-search"
 
+    def test_list_skills_by_activation_ordered_by_priority(self, db):
+        """Results are ordered by priority ascending, then name."""
+        _create_template(db, "s1", "style", "S", activation="default", priority=20)
+        _create_template(db, "s2", "safety", "F", activation="default", priority=10)
+        _create_template(db, "s3", "tone", "T", activation="default", priority=10)
+        results = db.list_skills_by_activation("default")
+        assert len(results) == 3
+        assert results[0]["name"] == "safety"
+        assert results[1]["name"] == "tone"
+        assert results[2]["name"] == "style"
+
+    def test_priority_default_is_zero(self, db):
+        """Priority defaults to 0 when not specified."""
+        _create_template(db, "s1", "skill", "content")
+        tpl = db.get_prompt_template("s1")
+        assert tpl is not None
+        assert tpl["priority"] == 0
+
+    def test_priority_roundtrip(self, db):
+        """Priority can be set on create and retrieved."""
+        _create_template(db, "s1", "skill", "content", priority=42)
+        tpl = db.get_prompt_template("s1")
+        assert tpl is not None
+        assert tpl["priority"] == 42
+
+    def test_priority_update(self, db):
+        """Priority can be updated."""
+        _create_template(db, "s1", "skill", "content", priority=10)
+        db.update_prompt_template("s1", priority=99)
+        tpl = db.get_prompt_template("s1")
+        assert tpl is not None
+        assert tpl["priority"] == 99
+
+    def test_list_default_templates_ordered_by_priority(self, db):
+        """list_default_templates() respects priority ordering."""
+        _create_template(db, "s1", "beta", "b", activation="default", priority=10)
+        _create_template(db, "s2", "alpha", "a", activation="default", priority=5)
+        _create_template(db, "s3", "gamma", "g", activation="default", priority=1)
+
+        results = db.list_default_templates()
+        assert len(results) == 3
+        assert results[0]["name"] == "gamma"
+        assert results[1]["name"] == "alpha"
+        assert results[2]["name"] == "beta"
+
 
 # ---------------------------------------------------------------------------
 # 1b. Skill resource storage tests
@@ -802,8 +848,8 @@ class TestSkillAPI:
         )
         assert resp.status_code == 404
 
-    def test_update_skill_readonly_rejected(self, api_client, api_storage):
-        """Updating a readonly (MCP-sourced) skill returns 403."""
+    def test_update_skill_readonly_spec_fields_rejected(self, api_client, api_storage):
+        """Updating spec fields on a readonly skill returns 400 (filtered to nothing)."""
         _create_template(
             api_storage,
             "s1",
@@ -815,9 +861,65 @@ class TestSkillAPI:
         )
         resp = api_client.put(
             "/v1/api/admin/skills/s1",
-            json={"description": "hacked"},
+            json={"description": "hacked", "content": "evil"},
         )
-        assert resp.status_code == 403
+        assert resp.status_code == 400
+        assert "runtime config" in resp.json()["error"].lower()
+
+    def test_update_skill_readonly_runtime_config_allowed(self, api_client, api_storage):
+        """Runtime config fields can be updated on a readonly (installed) skill."""
+        _create_template(
+            api_storage,
+            "s1",
+            "installed-skill",
+            "external content",
+            origin="source",
+            readonly=True,
+        )
+        resp = api_client.put(
+            "/v1/api/admin/skills/s1",
+            json={"model": "gpt-5", "temperature": 0.5, "enabled": False},
+        )
+        assert resp.status_code == 200
+        data = resp.json()
+        assert data["model"] == "gpt-5"
+        assert data["temperature"] == 0.5
+        assert data["enabled"] is False
+        # Spec fields must remain unchanged
+        assert data["content"] == "external content"
+
+    def test_update_skill_readonly_mixed_body_filters_spec(self, api_client, api_storage):
+        """When JS sends all fields for a readonly skill, spec fields are silently dropped."""
+        _create_template(
+            api_storage,
+            "s1",
+            "installed",
+            "original content",
+            origin="source",
+            readonly=True,
+        )
+        resp = api_client.put(
+            "/v1/api/admin/skills/s1",
+            # Simulate what the browser form submits: every field present
+            json={
+                "name": "hacked",
+                "content": "evil content",
+                "description": "tampered",
+                "model": "gpt-5",
+                "enabled": False,
+                "token_budget": 50000,
+            },
+        )
+        assert resp.status_code == 200
+        data = resp.json()
+        # Config fields updated
+        assert data["model"] == "gpt-5"
+        assert data["enabled"] is False
+        assert data["token_budget"] == 50000
+        # Spec fields unchanged
+        assert data["name"] == "installed"
+        assert data["content"] == "original content"
+        assert data["description"] == ""
 
     def test_update_skill_recomputes_token_estimate(self, api_client, api_storage):
         """Updating content recomputes token_estimate."""
@@ -1156,6 +1258,49 @@ class TestSkillSessionConfigApplication:
         parsed_tools = json.loads(tpl["allowed_tools"])
         assert parsed_tools == ["bash", "read_file", "write_file"]
 
+    def test_license_compatibility_roundtrip(self, db):
+        """Agent Skills spec fields license and compatibility round-trip."""
+        db.create_prompt_template(
+            template_id="spec1",
+            name="spec-fields-skill",
+            category="general",
+            content="Spec test.",
+            skill_license="Apache-2.0",
+            compatibility="Requires git, docker, jq",
+        )
+        tpl = db.get_skill_by_name("spec-fields-skill")
+        assert tpl is not None
+        assert tpl["license"] == "Apache-2.0"
+        assert tpl["compatibility"] == "Requires git, docker, jq"
+
+    def test_license_compatibility_default_empty(self, db):
+        """license and compatibility default to empty string."""
+        db.create_prompt_template(
+            template_id="spec2",
+            name="no-spec-fields",
+            category="general",
+            content="No spec fields.",
+        )
+        tpl = db.get_skill_by_name("no-spec-fields")
+        assert tpl is not None
+        assert tpl["license"] == ""
+        assert tpl["compatibility"] == ""
+
+    def test_update_license_compatibility(self, db):
+        """license and compatibility can be updated."""
+        db.create_prompt_template(
+            template_id="spec3",
+            name="updatable-spec",
+            category="general",
+            content="Test.",
+        )
+        db.update_prompt_template("spec3", license="MIT")
+        db.update_prompt_template("spec3", compatibility="Python 3.11+")
+        tpl = db.get_prompt_template("spec3")
+        assert tpl is not None
+        assert tpl["license"] == "MIT"
+        assert tpl["compatibility"] == "Python 3.11+"
+
 
 # ---------------------------------------------------------------------------
 # 7. Migration behavior tests
@@ -1456,3 +1601,322 @@ class TestSkillAdminEndpoints:
         )
         assert resp.status_code == 400
         assert "integer" in resp.json()["error"].lower()
+
+
+# ---------------------------------------------------------------------------
+# 9. Skill session config applied to workstream via server handler
+# ---------------------------------------------------------------------------
+
+
+class TestSkillConfigAppliedToWorkstream:
+    """Verify that skill session config fields are applied to the ChatSession
+    when a workstream is created via the server ``create_workstream`` handler.
+    """
+
+    @pytest.fixture()
+    def _ws_app(self, tmp_path):
+        """Build a minimal Starlette app with the real ``create_workstream``
+        handler, a real ``WorkstreamManager``, and a temp SQLite storage
+        backend.  Returns ``(TestClient, WorkstreamManager, storage)``.
+        """
+        import queue
+        import threading
+
+        import turnstone.core.storage._registry as _reg
+        from turnstone.core.workstream import WorkstreamManager
+        from turnstone.server import create_workstream
+
+        storage = SQLiteBackend(str(tmp_path / "ws_test.db"))
+
+        # Inject the test storage as the global singleton so that
+        # get_storage() / get_skill_by_name() resolve against it.
+        old_storage = _reg._storage
+        _reg._storage = storage
+
+        def _session_factory(
+            ui: Any, model_alias: Any = None, ws_id: Any = None, **kwargs: Any
+        ) -> ChatSession:
+            return ChatSession(
+                client=MagicMock(),
+                model=model_alias or "test-model",
+                ui=ui,
+                instructions=None,
+                temperature=0.5,
+                max_tokens=4096,
+                tool_timeout=30,
+                ws_id=ws_id,
+                skill=kwargs.get("skill"),
+            )
+
+        mgr = WorkstreamManager(_session_factory)
+
+        routes = [
+            Mount(
+                "/v1",
+                routes=[
+                    Route(
+                        "/api/workstreams/new",
+                        create_workstream,
+                        methods=["POST"],
+                    ),
+                ],
+            ),
+        ]
+        app = Starlette(
+            routes=routes,
+            middleware=[Middleware(_InjectAuthMiddleware)],
+        )
+        app.state.workstreams = mgr
+        app.state.skip_permissions = True
+        app.state.global_queue = queue.Queue()
+        app.state.global_listeners = []
+        app.state.global_listeners_lock = threading.Lock()
+
+        client = TestClient(app, raise_server_exceptions=False)
+
+        yield client, mgr, storage
+
+        # Restore original storage singleton.
+        _reg._storage = old_storage
+
+    def test_session_receives_temperature(self, _ws_app):
+        """Skill temperature overrides the session default."""
+        client, mgr, storage = _ws_app
+        _create_template(storage, "s1", "warm-skill", "Be warm.", temperature=0.9, enabled=True)
+
+        resp = client.post("/v1/api/workstreams/new", json={"skill": "warm-skill"})
+        assert resp.status_code == 200
+        ws_id = resp.json()["ws_id"]
+        ws = mgr.get(ws_id)
+        assert ws is not None and ws.session is not None
+        assert ws.session.temperature == 0.9
+
+    def test_session_receives_max_tokens(self, _ws_app):
+        """Skill max_tokens overrides the session default."""
+        client, mgr, storage = _ws_app
+        _create_template(storage, "s1", "token-skill", "Be concise.", max_tokens=1024, enabled=True)
+
+        resp = client.post("/v1/api/workstreams/new", json={"skill": "token-skill"})
+        assert resp.status_code == 200
+        ws = mgr.get(resp.json()["ws_id"])
+        assert ws is not None and ws.session is not None
+        assert ws.session.max_tokens == 1024
+
+    def test_session_receives_token_budget(self, _ws_app):
+        """Skill token_budget is applied to the session."""
+        client, mgr, storage = _ws_app
+        _create_template(
+            storage, "s1", "budget-skill", "Stay on budget.", token_budget=50000, enabled=True
+        )
+
+        resp = client.post("/v1/api/workstreams/new", json={"skill": "budget-skill"})
+        assert resp.status_code == 200
+        ws = mgr.get(resp.json()["ws_id"])
+        assert ws is not None and ws.session is not None
+        assert ws.session._token_budget == 50000
+
+    def test_session_receives_reasoning_effort(self, _ws_app):
+        """Skill reasoning_effort is applied to the session."""
+        client, mgr, storage = _ws_app
+        _create_template(
+            storage, "s1", "effort-skill", "Think hard.", reasoning_effort="high", enabled=True
+        )
+
+        resp = client.post("/v1/api/workstreams/new", json={"skill": "effort-skill"})
+        assert resp.status_code == 200
+        ws = mgr.get(resp.json()["ws_id"])
+        assert ws is not None and ws.session is not None
+        assert ws.session.reasoning_effort == "high"
+
+    def test_session_receives_agent_max_turns(self, _ws_app):
+        """Skill agent_max_turns is applied to the session."""
+        client, mgr, storage = _ws_app
+        _create_template(
+            storage, "s1", "turns-skill", "Few turns.", agent_max_turns=3, enabled=True
+        )
+
+        resp = client.post("/v1/api/workstreams/new", json={"skill": "turns-skill"})
+        assert resp.status_code == 200
+        ws = mgr.get(resp.json()["ws_id"])
+        assert ws is not None and ws.session is not None
+        assert ws.session.agent_max_turns == 3
+
+    def test_auto_approve_set_on_ui(self, _ws_app):
+        """Skill auto_approve=True propagates to the WebUI."""
+        from turnstone.server import WebUI
+
+        client, mgr, storage = _ws_app
+        _create_template(
+            storage, "s1", "approve-skill", "Auto approve.", auto_approve=True, enabled=True
+        )
+
+        resp = client.post("/v1/api/workstreams/new", json={"skill": "approve-skill"})
+        assert resp.status_code == 200
+        ws = mgr.get(resp.json()["ws_id"])
+        assert ws is not None
+        assert isinstance(ws.ui, WebUI)
+        assert ws.ui.auto_approve is True
+
+    def test_allowed_tools_set_on_ui(self, _ws_app):
+        """Skill allowed_tools are parsed and set as auto_approve_tools on the UI."""
+        from turnstone.server import WebUI
+
+        client, mgr, storage = _ws_app
+        _create_template(
+            storage,
+            "s1",
+            "tools-skill",
+            "Restricted tools.",
+            allowed_tools='["bash", "read_file"]',
+            enabled=True,
+        )
+
+        resp = client.post("/v1/api/workstreams/new", json={"skill": "tools-skill"})
+        assert resp.status_code == 200
+        ws = mgr.get(resp.json()["ws_id"])
+        assert ws is not None
+        assert isinstance(ws.ui, WebUI)
+        assert ws.ui.auto_approve_tools == {"bash", "read_file"}
+
+    def test_skill_model_overrides_resolved_model(self, _ws_app):
+        """Skill model field overrides the default session model."""
+        client, mgr, storage = _ws_app
+        _create_template(
+            storage, "s1", "model-skill", "Use specific model.", model="gpt-5", enabled=True
+        )
+
+        resp = client.post("/v1/api/workstreams/new", json={"skill": "model-skill"})
+        assert resp.status_code == 200
+        ws = mgr.get(resp.json()["ws_id"])
+        assert ws is not None and ws.session is not None
+        assert ws.session.model == "gpt-5"
+
+    def test_all_session_config_fields_applied(self, _ws_app):
+        """All session config fields from a skill are applied together."""
+        from turnstone.server import WebUI
+
+        client, mgr, storage = _ws_app
+        _create_template(
+            storage,
+            "s1",
+            "full-skill",
+            "Full config skill.",
+            model="gpt-5",
+            temperature=0.8,
+            reasoning_effort="high",
+            max_tokens=2048,
+            token_budget=100000,
+            agent_max_turns=10,
+            auto_approve=True,
+            allowed_tools='["bash", "write_file", "read_file"]',
+            notify_on_complete='{"channel": "discord"}',
+            enabled=True,
+        )
+
+        resp = client.post("/v1/api/workstreams/new", json={"skill": "full-skill"})
+        assert resp.status_code == 200
+        ws = mgr.get(resp.json()["ws_id"])
+        assert ws is not None and ws.session is not None
+        sess = ws.session
+        assert sess.model == "gpt-5"
+        assert sess.temperature == 0.8
+        assert sess.reasoning_effort == "high"
+        assert sess.max_tokens == 2048
+        assert sess._token_budget == 100000
+        assert sess.agent_max_turns == 10
+        assert sess._notify_on_complete == '{"channel": "discord"}'
+        assert sess._applied_skill_id == "s1"
+        assert sess._applied_skill_content == "Full config skill."
+        assert isinstance(ws.ui, WebUI)
+        assert ws.ui.auto_approve is True
+        assert ws.ui.auto_approve_tools == {"bash", "write_file", "read_file"}
+
+    def test_disabled_skill_returns_400(self, _ws_app):
+        """Creating a workstream with a disabled skill returns 400."""
+        client, _mgr, storage = _ws_app
+        _create_template(storage, "s1", "disabled-skill", "Disabled.", enabled=False)
+
+        resp = client.post("/v1/api/workstreams/new", json={"skill": "disabled-skill"})
+        assert resp.status_code == 400
+        assert "disabled" in resp.json()["error"].lower()
+
+    def test_unknown_skill_returns_400(self, _ws_app):
+        """Creating a workstream with a nonexistent skill returns 400."""
+        client, _mgr, _storage = _ws_app
+
+        resp = client.post("/v1/api/workstreams/new", json={"skill": "no-such-skill"})
+        assert resp.status_code == 400
+        assert "not found" in resp.json()["error"].lower()
+
+    def test_zero_token_budget_is_noop(self, _ws_app):
+        """Skill with token_budget=0 — handler skips budget application (> 0 guard)."""
+        client, mgr, storage = _ws_app
+        _create_template(
+            storage, "s1", "no-budget-skill", "No budget.", token_budget=0, enabled=True
+        )
+
+        resp = client.post("/v1/api/workstreams/new", json={"skill": "no-budget-skill"})
+        assert resp.status_code == 200
+        ws = mgr.get(resp.json()["ws_id"])
+        assert ws is not None and ws.session is not None
+        # Budget stays at default (0) — the handler's > 0 guard prevents application
+        assert ws.session._token_budget == 0
+
+    def test_empty_allowed_tools_is_noop(self, _ws_app):
+        """Skill with allowed_tools='[]' — handler skips (empty check)."""
+        client, mgr, storage = _ws_app
+        _create_template(
+            storage, "s1", "no-tools-skill", "No tools.", allowed_tools="[]", enabled=True
+        )
+
+        resp = client.post("/v1/api/workstreams/new", json={"skill": "no-tools-skill"})
+        assert resp.status_code == 200
+        ws = mgr.get(resp.json()["ws_id"])
+        assert ws is not None
+        # auto_approve_tools stays at default (empty set)
+        assert ws.ui.auto_approve_tools == set()
+
+    def test_skill_lineage_in_workstreams_table(self, _ws_app):
+        """skill_id and skill_version columns are populated in the workstreams table."""
+        import sqlalchemy as sa
+
+        from turnstone.core.storage._schema import workstreams
+
+        client, _mgr, storage = _ws_app
+        _create_template(storage, "s1", "lineage-skill", "Track me.", enabled=True)
+
+        resp = client.post("/v1/api/workstreams/new", json={"skill": "lineage-skill"})
+        assert resp.status_code == 200
+        ws_id = resp.json()["ws_id"]
+
+        with storage._engine.connect() as conn:
+            row = conn.execute(
+                sa.select(workstreams.c.skill_id, workstreams.c.skill_version).where(
+                    workstreams.c.ws_id == ws_id
+                )
+            ).fetchone()
+        assert row is not None
+        assert row[0] == "s1"
+        assert row[1] == 1
+
+    def test_no_skill_lineage_when_no_skill(self, _ws_app):
+        """Workstream without a skill has empty skill_id and zero skill_version."""
+        import sqlalchemy as sa
+
+        from turnstone.core.storage._schema import workstreams
+
+        client, _mgr, storage = _ws_app
+
+        resp = client.post("/v1/api/workstreams/new", json={})
+        assert resp.status_code == 200
+        ws_id = resp.json()["ws_id"]
+
+        with storage._engine.connect() as conn:
+            row = conn.execute(
+                sa.select(workstreams.c.skill_id, workstreams.c.skill_version).where(
+                    workstreams.c.ws_id == ws_id
+                )
+            ).fetchone()
+        assert row is not None
+        assert row[0] == ""
+        assert row[1] == 0

tests/test_storage_sqlite.py

@@ -1,18 +1,8 @@
 """Tests for the SQLite storage backend."""
 
-import pytest
-
-from turnstone.core.storage import init_storage, reset_storage
-
-
-@pytest.fixture
-def backend(tmp_path):
-    """Create a fresh SQLiteBackend for each test."""
-    reset_storage()
-    b = init_storage("sqlite", path=str(tmp_path / "test.db"), run_migrations=False)
-    yield b
-    reset_storage()
+from __future__ import annotations
 
+from typing import Any
 
 # -- Workstream registration ---------------------------------------------------
 
@@ -289,6 +279,73 @@ class TestWorkstreams:
         assert rows[0][6] == "node-a"
 
 
+# -- Structured memory touch ---------------------------------------------------
+
+
+class TestTouchStructuredMemory:
+    @staticmethod
+    def _create_memory(
+        backend: Any, name: str = "m1", scope: str = "global", scope_id: str = ""
+    ) -> None:
+        import uuid
+
+        backend.create_structured_memory(
+            memory_id=str(uuid.uuid4()),
+            name=name,
+            description="test desc",
+            mem_type="project",
+            scope=scope,
+            scope_id=scope_id,
+            content="test content",
+        )
+
+    def test_batch_touch_multiple(self, backend):
+        self._create_memory(backend, name="a")
+        self._create_memory(backend, name="b")
+        self._create_memory(backend, name="c")
+
+        count = backend.touch_structured_memories(
+            [
+                ("a", "global", ""),
+                ("b", "global", ""),
+                ("c", "global", ""),
+            ]
+        )
+        assert count == 3
+
+        for name in ("a", "b", "c"):
+            mem = backend.get_structured_memory_by_name(name, "global", "")
+            assert int(mem["access_count"]) == 1
+
+    def test_batch_touch_empty_list(self, backend):
+        assert backend.touch_structured_memories([]) == 0
+
+    def test_batch_touch_partial_match(self, backend):
+        self._create_memory(backend, name="exists")
+
+        count = backend.touch_structured_memories(
+            [
+                ("exists", "global", ""),
+                ("missing", "global", ""),
+            ]
+        )
+        assert count == 1
+
+        mem = backend.get_structured_memory_by_name("exists", "global", "")
+        assert int(mem["access_count"]) == 1
+
+    def test_batch_touch_with_duplicates(self, backend):
+        """Duplicate keys in batch should each increment access_count once."""
+        self._create_memory(backend, name="dup")
+
+        # Two identical keys — storage gets called twice for the same row
+        count = backend.touch_structured_memories([("dup", "global", ""), ("dup", "global", "")])
+        assert count == 2
+
+        mem = backend.get_structured_memory_by_name("dup", "global", "")
+        assert int(mem["access_count"]) == 2
+
+
 # -- Lifecycle -----------------------------------------------------------------
 
 

tests/test_structured_memory_storage.py

@@ -1,14 +1,5 @@
 """Tests for structured memory storage backend operations."""
 
-import pytest
-
-from turnstone.core.storage._sqlite import SQLiteBackend
-
-
-@pytest.fixture
-def backend(tmp_path):
-    return SQLiteBackend(str(tmp_path / "test.db"))
-
 
 class TestCreateAndGet:
     def test_create_and_get_by_id(self, backend):

tests/test_tls_admin.py

@@ -0,0 +1,234 @@
+"""Tests for TLS admin API endpoints and CLI commands."""
+
+from __future__ import annotations
+
+import pytest
+
+from turnstone.core.storage import get_storage, init_storage, reset_storage
+
+lacme = pytest.importorskip("lacme")
+
+
+@pytest.fixture(autouse=True)
+def _storage(tmp_path):
+    """Initialize ephemeral SQLite storage for each test."""
+    reset_storage()
+    db = str(tmp_path / "test.db")
+    init_storage("sqlite", path=db)
+    yield
+    reset_storage()
+
+
+@pytest.fixture
+def tls_manager():
+    """Create an initialized TLSManager."""
+    import asyncio
+
+    from turnstone.console.tls import TLSManager
+
+    mgr = TLSManager(get_storage())
+    asyncio.run(mgr.init_ca())
+    # Issue a test cert
+    asyncio.run(mgr.issue_console_certs(["test.internal", "localhost"]))
+    return mgr
+
+
+# ── Admin API endpoints ───────────────────────────────────────────────────────
+
+
+def _make_app(tls_manager):
+    """Create a minimal Starlette app with TLS endpoints."""
+    from starlette.applications import Starlette
+    from starlette.middleware import Middleware
+    from starlette.middleware.base import BaseHTTPMiddleware
+    from starlette.routing import Route
+
+    from turnstone.console.server import (
+        tls_ca_cert,
+        tls_ca_status,
+        tls_delete_cert,
+        tls_list_certs,
+        tls_renew_cert,
+    )
+    from turnstone.core.auth import AuthResult
+
+    async def _grant_access(request, call_next):  # type: ignore[no-untyped-def]
+        request.state.auth_result = AuthResult(
+            user_id="",
+            scopes=frozenset({"approve"}),
+            token_source="config",
+        )
+        return await call_next(request)
+
+    app = Starlette(
+        routes=[
+            Route("/ca", tls_ca_status),
+            Route("/ca.pem", tls_ca_cert),
+            Route("/certs", tls_list_certs),
+            Route("/certs/{domain}/renew", tls_renew_cert, methods=["POST"]),
+            Route("/certs/{domain}", tls_delete_cert, methods=["DELETE"]),
+        ],
+        middleware=[Middleware(BaseHTTPMiddleware, dispatch=_grant_access)],
+    )
+    app.state.tls_manager = tls_manager
+    return app
+
+
+def test_list_certs(tls_manager):
+    from starlette.testclient import TestClient
+
+    client = TestClient(_make_app(tls_manager))
+    resp = client.get("/certs")
+    assert resp.status_code == 200
+    data = resp.json()
+    assert len(data["certs"]) >= 1
+    assert data["certs"][0]["domain"] == "test.internal"
+
+
+def test_renew_cert(tls_manager):
+    from starlette.testclient import TestClient
+
+    client = TestClient(_make_app(tls_manager))
+    resp = client.post("/certs/test.internal/renew")
+    assert resp.status_code == 200
+    data = resp.json()
+    assert data["domain"] == "test.internal"
+
+
+def test_renew_cert_not_found(tls_manager):
+    from starlette.testclient import TestClient
+
+    client = TestClient(_make_app(tls_manager))
+    resp = client.post("/certs/nonexistent.internal/renew")
+    assert resp.status_code == 404
+
+
+def test_delete_cert(tls_manager):
+    from starlette.testclient import TestClient
+
+    client = TestClient(_make_app(tls_manager))
+    resp = client.delete("/certs/test.internal")
+    assert resp.status_code == 200
+    assert resp.json()["deleted"] == "test.internal"
+    # Verify it's gone
+    resp = client.get("/certs")
+    domains = [c["domain"] for c in resp.json()["certs"]]
+    assert "test.internal" not in domains
+
+
+def test_delete_cert_not_found(tls_manager):
+    from starlette.testclient import TestClient
+
+    client = TestClient(_make_app(tls_manager))
+    resp = client.delete("/certs/nonexistent.internal")
+    assert resp.status_code == 404
+
+
+# ── CLI bootstrap ─────────────────────────────────────────────────────────────
+
+
+def test_cli_bootstrap(tmp_path):
+    """Test offline CA bootstrap."""
+    import argparse
+
+    from turnstone.admin import _cmd_tls_bootstrap
+
+    out = tmp_path / "certs"
+    args = argparse.Namespace(out=str(out), issue=["redis.internal", "pg.internal"])
+    _cmd_tls_bootstrap(args)
+
+    assert (out / "ca.pem").exists()
+    assert b"BEGIN CERTIFICATE" in (out / "ca.pem").read_bytes()
+    # Check certs were issued
+    assert (out / "certs" / "redis.internal").exists()
+    assert (out / "certs" / "pg.internal").exists()
+
+
+def test_cli_bootstrap_no_issue(tmp_path):
+    """Bootstrap with no --issue creates CA only."""
+    import argparse
+
+    from turnstone.admin import _cmd_tls_bootstrap
+
+    out = tmp_path / "certs"
+    args = argparse.Namespace(out=str(out), issue=[])
+    _cmd_tls_bootstrap(args)
+
+    assert (out / "ca.pem").exists()
+    # No certs dir
+    certs_dir = out / "certs"
+    if certs_dir.exists():
+        assert len(list(certs_dir.iterdir())) == 0
+
+
+# ── Config parsing ────────────────────────────────────────────────────────────
+
+
+def test_redis_tls_config_map():
+    """Redis TLS keys are in the config map."""
+    from turnstone.core.config import _CONFIG_MAP
+
+    redis_map = _CONFIG_MAP["redis"]
+    assert "tls" in redis_map
+    assert "tls_ca" in redis_map
+    assert "tls_cert" in redis_map
+    assert "tls_key" in redis_map
+
+
+def test_database_ssl_config_map():
+    """Database SSL keys are in the config map."""
+    from turnstone.core.config import _CONFIG_MAP
+
+    db_map = _CONFIG_MAP["database"]
+    assert "sslmode" in db_map
+    assert "sslrootcert" in db_map
+    assert "sslcert" in db_map
+    assert "sslkey" in db_map
+
+
+# ── Auth enforcement ──────────────────────────────────────────────────────────
+
+
+def test_tls_endpoints_require_auth(tls_manager):
+    """TLS admin endpoints return 401 without auth."""
+    from starlette.applications import Starlette
+    from starlette.routing import Route
+    from starlette.testclient import TestClient
+
+    from turnstone.console.server import tls_ca_status, tls_list_certs
+
+    # No auth middleware — request.state.auth_result will be missing
+    app = Starlette(
+        routes=[
+            Route("/ca", tls_ca_status),
+            Route("/certs", tls_list_certs),
+        ]
+    )
+    app.state.tls_manager = tls_manager
+
+    client = TestClient(app)
+    resp = client.get("/ca")
+    assert resp.status_code == 401
+
+    resp = client.get("/certs")
+    assert resp.status_code == 401
+
+
+# ── SDK TLS params ────────────────────────────────────────────────────────────
+
+
+def test_sdk_client_cert_requires_both():
+    """SDK raises ValueError if only one of client_cert/client_key provided."""
+    from turnstone.sdk._base import _BaseClient
+
+    with pytest.raises(ValueError, match="Both client_cert and client_key"):
+        _BaseClient(
+            base_url="http://localhost:8080",
+            client_cert="/path/to/cert.pem",
+        )
+
+    with pytest.raises(ValueError, match="Both client_cert and client_key"):
+        _BaseClient(
+            base_url="http://localhost:8080",
+            client_key="/path/to/key.pem",
+        )

tests/test_tls_client.py

@@ -0,0 +1,98 @@
+"""Tests for TLSClient — service node certificate provisioning."""
+
+from __future__ import annotations
+
+from unittest.mock import MagicMock
+
+import pytest
+
+from turnstone.core.storage import get_storage, init_storage, reset_storage
+
+lacme = pytest.importorskip("lacme")
+
+
+@pytest.fixture(autouse=True)
+def _storage(tmp_path):
+    """Initialize ephemeral SQLite storage for each test."""
+    reset_storage()
+    db = str(tmp_path / "test.db")
+    init_storage("sqlite", path=db)
+    yield
+    reset_storage()
+
+
+# ── Console URL discovery ─────────────────────────────────────────────────────
+
+
+def test_discover_console_url():
+    """TLSClient discovers console URL from services table."""
+    from turnstone.core.tls import TLSClient
+
+    storage = get_storage()
+    storage.register_service("console", "console", "http://console:8080")
+
+    client = TLSClient(storage=storage, hostnames=["node-1"])
+    url = client._discover_console_url()
+    assert url == "http://console:8080"
+
+
+def test_discover_console_url_missing():
+    """TLSClient raises if no console registered."""
+    from turnstone.core.tls import TLSClient
+
+    client = TLSClient(storage=get_storage(), hostnames=["node-1"])
+    with pytest.raises(RuntimeError, match="No console service found"):
+        client._discover_console_url()
+
+
+def test_explicit_console_url_skips_discovery():
+    """When console_url is provided, discovery is skipped."""
+    from turnstone.core.tls import TLSClient
+
+    client = TLSClient(
+        storage=get_storage(),
+        console_url="http://explicit:9090",
+        hostnames=["node-1"],
+    )
+    assert client._console_url == "http://explicit:9090"
+
+
+# ── SSL context construction ─────────────────────────────────────────────────
+
+
+@pytest.mark.anyio
+async def test_ssl_contexts_none_before_init():
+    """SSL contexts are None before init()."""
+    from turnstone.core.tls import TLSClient
+
+    client = TLSClient(
+        storage=get_storage(),
+        console_url="http://localhost:8080",
+        hostnames=["node-1"],
+    )
+    assert client.get_server_ssl_context() is None
+    assert client.get_client_ssl_context() is None
+    assert not client.initialized
+
+
+# ── Backward compatibility ───────────────────────────────────────────────────
+
+
+def test_bridge_tls_defaults():
+    """Bridge with default TLS params works without changes."""
+    from turnstone.mq.bridge import Bridge
+
+    # Default: tls_verify=True, tls_cert=None — no mTLS
+    bridge = Bridge(server_url="http://localhost:8080")
+    assert bridge._tls_verify is True
+    assert bridge._tls_cert is None
+
+
+def test_collector_tls_defaults():
+    """Collector with default TLS params works without changes."""
+    from turnstone.console.collector import ClusterCollector
+
+    broker_mock = MagicMock()
+    collector = ClusterCollector(broker=broker_mock)
+    # Should create httpx client without errors
+    assert collector._http_client is not None

tests/test_tls_manager.py

@@ -0,0 +1,226 @@
+"""Tests for TLSManager — console CA and ACME server."""
+
+from __future__ import annotations
+
+import pytest
+
+from turnstone.core.storage import get_storage, init_storage, reset_storage
+
+lacme = pytest.importorskip("lacme")
+
+
+@pytest.fixture(autouse=True)
+def _storage(tmp_path):
+    """Initialize ephemeral SQLite storage for each test."""
+    reset_storage()
+    db = str(tmp_path / "test.db")
+    init_storage("sqlite", path=db)
+    yield
+    reset_storage()
+
+
+@pytest.fixture
+def tls_manager():
+    """Create a TLSManager backed by test storage."""
+    from turnstone.console.tls import TLSManager
+
+    return TLSManager(get_storage())
+
+
+# ── CA initialization ─────────────────────────────────────────────────────────
+
+
+@pytest.mark.anyio
+async def test_init_ca(tls_manager):
+    await tls_manager.init_ca()
+    assert tls_manager.ca_initialized
+    root_pem = tls_manager.get_root_cert_pem()
+    assert b"BEGIN CERTIFICATE" in root_pem
+
+
+@pytest.mark.anyio
+async def test_init_ca_persists(tls_manager):
+    """CA root survives re-initialization (loaded from storage)."""
+    await tls_manager.init_ca()
+    pem1 = tls_manager.get_root_cert_pem()
+
+    # Create a new manager on the same storage
+    from turnstone.console.tls import TLSManager
+
+    mgr2 = TLSManager(get_storage())
+    await mgr2.init_ca()
+    pem2 = mgr2.get_root_cert_pem()
+
+    assert pem1 == pem2  # Same CA loaded from DB
+
+
+@pytest.mark.anyio
+async def test_get_responder_before_init(tls_manager):
+    with pytest.raises(RuntimeError, match="CA not initialized"):
+        tls_manager.get_responder()
+
+
+@pytest.mark.anyio
+async def test_get_responder(tls_manager):
+    await tls_manager.init_ca()
+    responder = tls_manager.get_responder()
+    assert responder is not None
+    # Should be an ASGI app (callable)
+    assert callable(responder)
+
+
+# ── Cert issuance ─────────────────────────────────────────────────────────────
+
+
+@pytest.mark.anyio
+async def test_issue_console_certs_internal(tls_manager):
+    """Console certs issued from internal CA when no external directory."""
+    await tls_manager.init_ca()
+    await tls_manager.issue_console_certs(["console.internal", "localhost"])
+    assert tls_manager.internal_bundle is not None
+    assert tls_manager.frontend_bundle is not None
+    assert tls_manager.internal_bundle.domain == "console.internal"
+    assert b"BEGIN CERTIFICATE" in tls_manager.internal_bundle.cert_pem
+
+
+@pytest.mark.anyio
+async def test_issue_console_certs_persists(tls_manager):
+    """Certs loaded from storage on re-issue."""
+    await tls_manager.init_ca()
+    await tls_manager.issue_console_certs(["console.internal"])
+    bundle1 = tls_manager.internal_bundle
+
+    # New manager, same storage
+    from turnstone.console.tls import TLSManager
+
+    mgr2 = TLSManager(get_storage())
+    await mgr2.init_ca()
+    await mgr2.issue_console_certs(["console.internal"])
+    bundle2 = mgr2.internal_bundle
+
+    assert bundle1.cert_pem == bundle2.cert_pem
+
+
+# ── SSL contexts ──────────────────────────────────────────────────────────────
+
+
+@pytest.mark.anyio
+async def test_ssl_contexts_none_before_certs(tls_manager):
+    await tls_manager.init_ca()
+    assert tls_manager.get_server_ssl_context() is None
+    assert tls_manager.get_client_ssl_context() is None
+
+
+@pytest.mark.anyio
+async def test_ssl_contexts_after_certs(tls_manager):
+    await tls_manager.init_ca()
+    await tls_manager.issue_console_certs(["console.internal"])
+    server_ctx = tls_manager.get_server_ssl_context()
+    client_ctx = tls_manager.get_client_ssl_context()
+    assert server_ctx is not None
+    assert client_ctx is not None
+    import ssl
+
+    assert isinstance(server_ctx, ssl.SSLContext)
+    assert isinstance(client_ctx, ssl.SSLContext)
+
+
+# ── Root cert endpoint ────────────────────────────────────────────────────────
+
+
+@pytest.mark.anyio
+async def test_tls_ca_cert_endpoint(tls_manager):
+    """Test the CA cert download endpoint via test client."""
+    await tls_manager.init_ca()
+
+    from starlette.applications import Starlette
+    from starlette.middleware import Middleware
+    from starlette.routing import Route
+    from starlette.testclient import TestClient
+
+    from turnstone.console.server import tls_ca_cert, tls_ca_status
+
+    # Middleware that grants full access (config-token style: no user_id)
+    from turnstone.core.auth import AuthResult
+
+    async def _grant_access(request, call_next):  # type: ignore[no-untyped-def]
+        request.state.auth_result = AuthResult(
+            user_id="", scopes=frozenset({"approve"}), token_source="config"
+        )
+        return await call_next(request)
+
+    from starlette.middleware.base import BaseHTTPMiddleware
+
+    app = Starlette(
+        routes=[
+            Route("/ca.pem", tls_ca_cert),
+            Route("/ca", tls_ca_status),
+        ],
+        middleware=[Middleware(BaseHTTPMiddleware, dispatch=_grant_access)],
+    )
+    app.state.tls_manager = tls_manager
+
+    client = TestClient(app)
+
+    # CA cert download
+    resp = client.get("/ca.pem")
+    assert resp.status_code == 200
+    assert b"BEGIN CERTIFICATE" in resp.content
+    assert resp.headers["content-type"] == "application/x-pem-file"
+
+    # CA status
+    resp = client.get("/ca")
+    assert resp.status_code == 200
+    data = resp.json()
+    assert data["enabled"] is True
+    assert data["ca_cn"] == "Turnstone CA"
+
+
+@pytest.mark.anyio
+async def test_tls_endpoints_disabled():
+    """Endpoints return 404/disabled when TLS not enabled."""
+    from starlette.applications import Starlette
+    from starlette.middleware import Middleware
+    from starlette.middleware.base import BaseHTTPMiddleware
+    from starlette.routing import Route
+    from starlette.testclient import TestClient
+
+    from turnstone.console.server import tls_ca_cert, tls_ca_status
+    from turnstone.core.auth import AuthResult
+
+    async def _grant_access(request, call_next):  # type: ignore[no-untyped-def]
+        request.state.auth_result = AuthResult(
+            user_id="", scopes=frozenset({"approve"}), token_source="config"
+        )
+        return await call_next(request)
+
+    app = Starlette(
+        routes=[
+            Route("/ca.pem", tls_ca_cert),
+            Route("/ca", tls_ca_status),
+        ],
+        middleware=[Middleware(BaseHTTPMiddleware, dispatch=_grant_access)],
+    )
+    # No tls_manager on state
+
+    client = TestClient(app)
+
+    resp = client.get("/ca.pem")
+    assert resp.status_code == 404
+
+    resp = client.get("/ca")
+    data = resp.json()
+    assert data["enabled"] is False
+
+
+# ── Events ────────────────────────────────────────────────────────────────────
+
+
+@pytest.mark.anyio
+async def test_event_dispatcher_wired(tls_manager):
+    """Verify the event dispatcher has subscribers."""
+    assert tls_manager._event_dispatcher is not None
+    # Should have at least 4 subscriptions (issued, renewed, expiring, failed)
+    # The exact check depends on lacme's EventDispatcher internals,
+    # so just verify the dispatcher exists and the manager initializes cleanly
+    await tls_manager.init_ca()

tests/test_tls_storage.py

@@ -0,0 +1,226 @@
+"""Tests for TLS storage backend and lacme Store adapter."""
+
+from __future__ import annotations
+
+import json
+from datetime import UTC, datetime
+
+import pytest
+
+from turnstone.core.storage import get_storage, init_storage, reset_storage
+
+
+@pytest.fixture(autouse=True)
+def _storage(tmp_path):
+    """Initialize ephemeral SQLite storage for each test."""
+    reset_storage()
+    db = str(tmp_path / "test.db")
+    init_storage("sqlite", path=db)
+    yield
+    reset_storage()
+
+
+# ── Account keys ──────────────────────────────────────────────────────────────
+
+
+def test_save_and_load_account_key():
+    s = get_storage()
+    s.save_tls_account_key(
+        "default",
+        "-----BEGIN EC PRIVATE KEY-----\nfake\n-----END EC PRIVATE KEY-----",
+    )
+    result = s.load_tls_account_key("default")
+    assert result is not None
+    assert "EC PRIVATE KEY" in result
+
+
+def test_load_account_key_missing():
+    s = get_storage()
+    assert s.load_tls_account_key("nonexistent") is None
+
+
+def test_save_account_key_upsert():
+    s = get_storage()
+    s.save_tls_account_key("default", "key-v1")
+    s.save_tls_account_key("default", "key-v2")
+    assert s.load_tls_account_key("default") == "key-v2"
+
+
+# ── CA ────────────────────────────────────────────────────────────────────────
+
+
+def test_save_and_load_ca():
+    s = get_storage()
+    s.save_tls_ca("Turnstone CA", "cert-pem-data", "key-pem-data")
+    result = s.load_tls_ca("Turnstone CA")
+    assert result is not None
+    assert result["cert_pem"] == "cert-pem-data"
+    assert result["key_pem"] == "key-pem-data"
+    assert result["name"] == "Turnstone CA"
+
+
+def test_load_ca_missing():
+    s = get_storage()
+    assert s.load_tls_ca("nonexistent") is None
+
+
+def test_save_ca_upsert():
+    s = get_storage()
+    s.save_tls_ca("CA", "cert-v1", "key-v1")
+    s.save_tls_ca("CA", "cert-v2", "key-v2")
+    result = s.load_tls_ca("CA")
+    assert result["cert_pem"] == "cert-v2"
+    assert result["key_pem"] == "key-v2"
+
+
+# ── Certificates ──────────────────────────────────────────────────────────────
+
+
+def test_save_and_load_cert():
+    s = get_storage()
+    s.save_tls_cert(
+        domain="node-1.internal",
+        cert_pem="cert-data",
+        fullchain_pem="fullchain-data",
+        key_pem="key-data",
+        issued_at="2026-03-25T00:00:00",
+        expires_at="2026-03-27T00:00:00",
+        meta=json.dumps({"domains": ["node-1.internal", "10.0.1.5"]}),
+    )
+    result = s.load_tls_cert("node-1.internal")
+    assert result is not None
+    assert result["domain"] == "node-1.internal"
+    assert result["cert_pem"] == "cert-data"
+    assert result["fullchain_pem"] == "fullchain-data"
+    assert result["key_pem"] == "key-data"
+    assert result["issued_at"] == "2026-03-25T00:00:00"
+    assert result["expires_at"] == "2026-03-27T00:00:00"
+    meta = json.loads(result["meta"])
+    assert meta["domains"] == ["node-1.internal", "10.0.1.5"]
+
+
+def test_load_cert_missing():
+    s = get_storage()
+    assert s.load_tls_cert("nonexistent") is None
+
+
+def test_save_cert_upsert():
+    s = get_storage()
+    s.save_tls_cert("d", "c1", "f1", "k1", "2026-01-01", "2026-01-02")
+    s.save_tls_cert("d", "c2", "f2", "k2", "2026-02-01", "2026-02-02")
+    result = s.load_tls_cert("d")
+    assert result["cert_pem"] == "c2"
+    assert result["issued_at"] == "2026-02-01"
+
+
+def test_list_certs_empty():
+    s = get_storage()
+    assert s.list_tls_certs() == []
+
+
+def test_list_certs():
+    s = get_storage()
+    s.save_tls_cert("alpha.internal", "c", "f", "k", "2026-01-01", "2026-01-02")
+    s.save_tls_cert("beta.internal", "c", "f", "k", "2026-01-01", "2026-01-02")
+    certs = s.list_tls_certs()
+    assert len(certs) == 2
+    assert certs[0]["domain"] == "alpha.internal"  # sorted by domain
+    assert certs[1]["domain"] == "beta.internal"
+
+
+def test_delete_cert():
+    s = get_storage()
+    s.save_tls_cert("d", "c", "f", "k", "2026-01-01", "2026-01-02")
+    assert s.delete_tls_cert("d") is True
+    assert s.load_tls_cert("d") is None
+
+
+def test_delete_cert_missing():
+    s = get_storage()
+    assert s.delete_tls_cert("nonexistent") is False
+
+
+# ── StorageStore adapter ──────────────────────────────────────────────────────
+
+
+@pytest.fixture
+def store_adapter():
+    """Create a StorageStore backed by the test database."""
+    from turnstone.core.tls_store import StorageStore
+
+    return StorageStore(get_storage())
+
+
+def test_adapter_save_load_ca(store_adapter):
+    store_adapter.save_ca("test-ca", b"cert-pem", b"key-pem")
+    result = store_adapter.load_ca("test-ca")
+    assert result is not None
+    cert_pem, key_pem = result
+    assert cert_pem == b"cert-pem"
+    assert key_pem == b"key-pem"
+
+
+def test_adapter_load_ca_missing(store_adapter):
+    assert store_adapter.load_ca("missing") is None
+
+
+def test_adapter_save_load_cert(store_adapter):
+    lacme = pytest.importorskip("lacme")
+    now = datetime.now(UTC)
+    bundle = lacme.CertBundle(
+        domain="test.internal",
+        domains=("test.internal", "10.0.1.1"),
+        cert_pem=b"cert",
+        fullchain_pem=b"fullchain",
+        key_pem=b"key",
+        issued_at=now,
+        expires_at=now,
+    )
+    store_adapter.save_cert(bundle)
+    loaded = store_adapter.load_cert("test.internal")
+    assert loaded is not None
+    assert loaded.domain == "test.internal"
+    assert loaded.domains == ("test.internal", "10.0.1.1")
+    assert loaded.cert_pem == b"cert"
+    assert loaded.fullchain_pem == b"fullchain"
+    assert loaded.key_pem == b"key"
+
+
+def test_adapter_list_certs(store_adapter):
+    lacme = pytest.importorskip("lacme")
+    now = datetime.now(UTC)
+    for name in ["alpha", "beta"]:
+        bundle = lacme.CertBundle(
+            domain=f"{name}.internal",
+            domains=(f"{name}.internal",),
+            cert_pem=b"c",
+            fullchain_pem=b"f",
+            key_pem=b"k",
+            issued_at=now,
+            expires_at=now,
+        )
+        store_adapter.save_cert(bundle)
+    certs = store_adapter.list_certs()
+    assert len(certs) == 2
+    assert certs[0].domain == "alpha.internal"
+
+
+def test_adapter_load_cert_missing(store_adapter):
+    assert store_adapter.load_cert("missing") is None
+
+
+def test_adapter_account_key_roundtrip(store_adapter):
+    """Test account key save/load with real cryptography objects."""
+    pytest.importorskip("lacme")
+    from cryptography.hazmat.primitives.asymmetric import ec
+
+    key = ec.generate_private_key(ec.SECP256R1())
+    store_adapter.save_account_key(key)
+    loaded = store_adapter.load_account_key()
+    assert loaded is not None
+    # Verify it's a usable EC key
+    assert loaded.key_size == key.key_size
+
+
+def test_adapter_account_key_missing(store_adapter):
+    assert store_adapter.load_account_key() is None

tests/test_tool_policies_enforcement.py

@@ -0,0 +1,174 @@
+"""Tests for tool policy enforcement across CLI, bridge, and channel entry points."""
+
+from __future__ import annotations
+
+from unittest.mock import MagicMock, patch
+
+from turnstone.cli import TerminalUI
+
+# ---------------------------------------------------------------------------
+# CLI
+# ---------------------------------------------------------------------------
+
+
+class TestCLIPolicyEnforcement:
+    """Tool policies should be enforced in CLI approve_tools()."""
+
+    def _make_items(self, *tool_names: str) -> list[dict]:
+        return [
+            {
+                "call_id": f"call_{i}",
+                "header": f"Tool: {name}",
+                "preview": "",
+                "func_name": name,
+                "approval_label": name,
+                "needs_approval": True,
+            }
+            for i, name in enumerate(tool_names)
+        ]
+
+    def test_deny_policy_blocks_tool(self):
+        """A 'deny' policy verdict should block the tool without prompting."""
+        ui = TerminalUI()
+        items = self._make_items("bash")
+
+        with (
+            patch(
+                "turnstone.core.policy.evaluate_tool_policies_batch",
+                return_value={"bash": "deny"},
+            ),
+            patch(
+                "turnstone.core.storage._registry.get_storage",
+                return_value=MagicMock(),
+            ),
+        ):
+            approved, _ = ui.approve_tools(items)
+
+        assert items[0].get("denied") is True
+        assert items[0].get("error")
+        assert "policy" in items[0]["error"].lower()
+
+    def test_allow_policy_auto_approves(self):
+        """An 'allow' policy verdict should auto-approve without prompting."""
+        ui = TerminalUI()
+        items = self._make_items("read_file")
+
+        with (
+            patch(
+                "turnstone.core.policy.evaluate_tool_policies_batch",
+                return_value={"read_file": "allow"},
+            ),
+            patch(
+                "turnstone.core.storage._registry.get_storage",
+                return_value=MagicMock(),
+            ),
+        ):
+            approved, _ = ui.approve_tools(items)
+
+        assert approved is True
+
+    def test_no_storage_skips_policies(self):
+        """When storage is unavailable, policies are skipped (best-effort)."""
+        ui = TerminalUI()
+        items = self._make_items("bash")
+
+        with (
+            patch(
+                "turnstone.core.storage._registry.get_storage",
+                return_value=None,
+            ),
+            patch("builtins.input", return_value="y"),
+        ):
+            approved, _ = ui.approve_tools(items)
+
+        # Should fall through to normal prompt (which we answered 'y')
+        assert approved is True
+
+
+# ---------------------------------------------------------------------------
+# Bridge
+# ---------------------------------------------------------------------------
+
+
+class TestBridgePolicyEnforcement:
+    """Tool policies should be enforced in bridge _handle_approval()."""
+
+    def _make_bridge(self):
+        from turnstone.mq.bridge import Bridge
+
+        broker = MagicMock()
+        return Bridge(
+            server_url="http://localhost:8080",
+            broker=broker,
+            node_id="test-node",
+            approval_timeout=1,
+        )
+
+    def _approval_items(self, *tool_names: str) -> list[dict]:
+        return [
+            {"func_name": name, "needs_approval": True, "approval_label": name}
+            for name in tool_names
+        ]
+
+    def test_deny_policy_rejects_approval(self):
+        """A 'deny' policy should reject the approval."""
+        bridge = self._make_bridge()
+
+        with (
+            patch(
+                "turnstone.core.policy.evaluate_tool_policies_batch",
+                return_value={"bash": "deny"},
+            ),
+            patch(
+                "turnstone.core.storage._registry._storage",
+                new=MagicMock(),
+            ),
+            patch.object(bridge, "_api_approve") as mock_approve,
+            patch.object(bridge, "_publish_ws"),
+        ):
+            bridge._handle_approval("ws-1", {"items": self._approval_items("bash")})
+
+        mock_approve.assert_called_once()
+        assert mock_approve.call_args.kwargs.get("approved") is False
+
+    def test_allow_policy_approves(self):
+        """An 'allow' policy should auto-approve."""
+        bridge = self._make_bridge()
+
+        with (
+            patch(
+                "turnstone.core.policy.evaluate_tool_policies_batch",
+                return_value={"read_file": "allow"},
+            ),
+            patch(
+                "turnstone.core.storage._registry.get_storage",
+                return_value=MagicMock(),
+            ),
+            patch.object(bridge, "_api_approve") as mock_approve,
+            patch.object(bridge, "_publish_ws"),
+        ):
+            bridge._handle_approval("ws-1", {"items": self._approval_items("read_file")})
+
+        mock_approve.assert_called_once()
+        assert mock_approve.call_args.kwargs.get("approved") is True
+
+    def test_mixed_deny_rejects_batch(self):
+        """If any tool is denied, the whole batch is rejected."""
+        bridge = self._make_bridge()
+
+        with (
+            patch(
+                "turnstone.core.policy.evaluate_tool_policies_batch",
+                return_value={"bash": "deny", "read_file": "allow"},
+            ),
+            patch(
+                "turnstone.core.storage._registry._storage",
+                new=MagicMock(),
+            ),
+            patch.object(bridge, "_api_approve") as mock_approve,
+            patch.object(bridge, "_publish_ws"),
+        ):
+            bridge._handle_approval("ws-1", {"items": self._approval_items("bash", "read_file")})
+
+        mock_approve.assert_called_once()
+        assert mock_approve.call_args.kwargs.get("approved") is False

tests/test_tool_search.py

@@ -128,17 +128,9 @@ class TestToolSearchManager:
         return ToolSearchManager(
             all_tools,
             always_on_names={"bash", "read_file", "edit_file"},
-            threshold=5,
             max_results=3,
         )
 
-    def test_should_activate_above_threshold(self, manager):
-        assert manager.should_activate()
-
-    def test_should_not_activate_below_threshold(self, builtin_tools):
-        mgr = ToolSearchManager(builtin_tools, always_on_names={"bash", "read_file", "edit_file"})
-        assert not mgr.should_activate()
-
     def test_visible_tools_initially_builtin_only(self, manager):
         visible = manager.get_visible_tools()
         names = {_tool_name(t) for t in visible}
@@ -201,9 +193,6 @@ class TestToolSearchManager:
         names = {_tool_name(t) for t in deferred}
         assert "mcp__github__create_issue" not in names
 
-    def test_get_all_tools_returns_everything(self, manager, builtin_tools, mcp_tools):
-        assert len(manager.get_all_tools()) == len(builtin_tools) + len(mcp_tools)
-
     def test_search_tool_definition_format(self, manager):
         defn = manager.get_search_tool_definition()
         assert defn["type"] == "function"

tests/test_tools_schema.py

@@ -104,15 +104,15 @@ class TestToolsMetadata:
             "man": "page",
             "web_fetch": "url",
             "web_search": "query",
-            "task": "prompt",
-            "create_plan": "goal",
+            "task_agent": "prompt",
+            "plan_agent": "goal",
             "memory": "name",
             "recall": "query",
             "notify": "message",
             "watch": "command",
             "read_resource": "uri",
             "use_prompt": "name",
-            "load_skill": "name",
+            "skill": "name",
         }
         assert expected == PRIMARY_KEY_MAP
 

tests/test_user_storage.py

@@ -2,16 +2,6 @@
 
 from __future__ import annotations
 
-import pytest
-
-from turnstone.core.storage._sqlite import SQLiteBackend
-
-
-@pytest.fixture()
-def db(tmp_path):
-    """Create a fresh SQLite backend for each test."""
-    return SQLiteBackend(str(tmp_path / "test.db"))
-
 
 class TestUserCRUD:
     def test_create_and_get(self, db):

tests/test_watch_storage.py

@@ -2,16 +2,6 @@
 
 from __future__ import annotations
 
-import pytest
-
-from turnstone.core.storage._sqlite import SQLiteBackend
-
-
-@pytest.fixture
-def db(tmp_path):
-    """Fresh SQLite backend for each test."""
-    return SQLiteBackend(str(tmp_path / "test.db"))
-
 
 def _make_watch_kwargs(**overrides):
     """Build default kwargs for create_watch."""

tests/test_web_search.py

@@ -0,0 +1,175 @@
+"""Tests for turnstone.core.web_search — pluggable web search backends."""
+
+from __future__ import annotations
+
+from unittest.mock import MagicMock, patch
+
+from turnstone.core.web_search import (
+    DuckDuckGoClient,
+    MCPSearchClient,
+    TavilyClient,
+    _format_ddg,
+    _format_tavily,
+    resolve_web_search_client,
+)
+
+# ---------------------------------------------------------------------------
+# Formatters
+# ---------------------------------------------------------------------------
+
+
+class TestFormatTavily:
+    def test_formats_answer_and_results(self):
+        data = {
+            "answer": "Python is great",
+            "results": [
+                {"title": "Python.org", "url": "https://python.org", "content": "Official site"},
+                {"title": "PyPI", "url": "https://pypi.org", "content": "Package index"},
+            ],
+        }
+        out = _format_tavily(data, "python")
+        assert "Answer: Python is great" in out
+        assert "[Python.org](https://python.org)" in out
+        assert "[PyPI](https://pypi.org)" in out
+
+    def test_no_results(self):
+        out = _format_tavily({"results": []}, "nothing")
+        assert "No results for 'nothing'" in out
+
+    def test_no_answer(self):
+        data = {
+            "results": [{"title": "T", "url": "http://t", "content": "C"}],
+        }
+        out = _format_tavily(data, "q")
+        assert "Answer:" not in out
+        assert "[T](http://t)" in out
+
+
+class TestFormatDDG:
+    def test_formats_results(self):
+        results = [
+            {"title": "DDG Result", "href": "https://ddg.example.com", "body": "Search body"},
+        ]
+        out = _format_ddg(results, "test")
+        assert "[DDG Result](https://ddg.example.com)" in out
+        assert "Search body" in out
+
+    def test_no_results(self):
+        out = _format_ddg([], "nothing")
+        assert "No results for 'nothing'" in out
+
+
+# ---------------------------------------------------------------------------
+# Client tests
+# ---------------------------------------------------------------------------
+
+
+class TestTavilyClient:
+    def test_search_calls_api(self):
+        mock_resp = MagicMock()
+        mock_resp.json.return_value = {
+            "answer": "42",
+            "results": [{"title": "T", "url": "http://t", "content": "C"}],
+        }
+        with patch("turnstone.core.web_search.httpx.post", return_value=mock_resp) as mock_post:
+            client = TavilyClient("test-key", timeout=10)
+            result = client.search("meaning of life", max_results=3)
+
+        mock_post.assert_called_once()
+        call_kwargs = mock_post.call_args
+        assert call_kwargs.kwargs["json"]["query"] == "meaning of life"
+        assert call_kwargs.kwargs["json"]["max_results"] == 3
+        assert "Answer: 42" in result
+
+
+class TestDuckDuckGoClient:
+    def test_integration_via_mock_ddgs(self):
+        """Patch the ddgs import inside DuckDuckGoClient.search."""
+        mock_ddgs = MagicMock()
+        mock_ddgs.__enter__ = MagicMock(return_value=mock_ddgs)
+        mock_ddgs.__exit__ = MagicMock(return_value=False)
+        mock_ddgs.text.return_value = [
+            {"title": "DDG Result", "href": "https://ddg.co", "body": "Found it"},
+        ]
+        mock_module = MagicMock()
+        mock_module.DDGS.return_value = mock_ddgs
+        with patch.dict("sys.modules", {"ddgs": mock_module}):
+            client = DuckDuckGoClient(timeout=10)
+            result = client.search("test query", max_results=3)
+        mock_ddgs.text.assert_called_once_with("test query", max_results=3)
+        assert "[DDG Result](https://ddg.co)" in result
+        assert "Found it" in result
+
+
+class TestMCPSearchClient:
+    def test_delegates_to_mcp(self):
+        mcp = MagicMock()
+        mcp.call_tool_sync.return_value = "MCP search results"
+        client = MCPSearchClient(mcp, "mcp__ddg__search", timeout=30)
+        result = client.search("test", max_results=3, topic="news")
+        mcp.call_tool_sync.assert_called_once_with(
+            "mcp__ddg__search",
+            {"query": "test", "max_results": 3, "topic": "news"},
+            timeout=30,
+        )
+        assert result == "MCP search results"
+
+
+# ---------------------------------------------------------------------------
+# Resolver
+# ---------------------------------------------------------------------------
+
+
+class TestResolveClient:
+    def test_auto_tavily_when_key_present(self):
+        client = resolve_web_search_client("", tavily_key="key")
+        assert isinstance(client, TavilyClient)
+
+    def test_auto_ddg_when_no_tavily(self):
+        with patch("turnstone.core.web_search._ddg_available", return_value=True):
+            client = resolve_web_search_client("", tavily_key=None)
+        assert isinstance(client, DuckDuckGoClient)
+
+    def test_auto_none_when_nothing_available(self):
+        with patch("turnstone.core.web_search._ddg_available", return_value=False):
+            client = resolve_web_search_client("", tavily_key=None)
+        assert client is None
+
+    def test_explicit_tavily(self):
+        client = resolve_web_search_client("tavily", tavily_key="key")
+        assert isinstance(client, TavilyClient)
+
+    def test_explicit_tavily_no_key(self):
+        client = resolve_web_search_client("tavily", tavily_key=None)
+        assert client is None
+
+    def test_explicit_ddg(self):
+        with patch("turnstone.core.web_search._ddg_available", return_value=True):
+            client = resolve_web_search_client("ddg", tavily_key=None)
+        assert isinstance(client, DuckDuckGoClient)
+
+    def test_explicit_ddg_not_installed(self):
+        with patch("turnstone.core.web_search._ddg_available", return_value=False):
+            client = resolve_web_search_client("ddg", tavily_key=None)
+        assert client is None
+
+    def test_mcp_backend(self):
+        mcp = MagicMock()
+        mcp.is_mcp_tool.return_value = True
+        client = resolve_web_search_client("mcp:ddg:search", tavily_key=None, mcp_client=mcp)
+        assert isinstance(client, MCPSearchClient)
+        mcp.is_mcp_tool.assert_called_with("mcp__ddg__search")
+
+    def test_mcp_backend_not_connected(self):
+        mcp = MagicMock()
+        mcp.is_mcp_tool.return_value = False
+        client = resolve_web_search_client("mcp:ddg:search", tavily_key=None, mcp_client=mcp)
+        assert client is None
+
+    def test_mcp_backend_no_client(self):
+        client = resolve_web_search_client("mcp:ddg:search", tavily_key=None, mcp_client=None)
+        assert client is None
+
+    def test_unknown_backend_returns_none(self):
+        client = resolve_web_search_client("typo_backend", tavily_key="key")
+        assert client is None

turnstone.example.toml

@@ -0,0 +1,124 @@
+# turnstone.toml — shared bootstrap configuration
+#
+# This file is read once at startup. Values here are overridden by
+# environment variables, which are in turn overridden by CLI flags.
+#
+# All sections are optional. Missing sections use binary defaults.
+# Config file location precedence:
+#   1. --config flag
+#   2. $TURNSTONE_CONFIG env var
+#   3. ~/.config/turnstone/config.toml
+
+# --- LLM API (turnstone, node, eval) ---
+
+[api]
+# base_url = ""                # API endpoint; empty = binary default
+# api_key = ""                 # env: OPENAI_API_KEY or ANTHROPIC_API_KEY
+
+# --- Default Model (turnstone, node, eval) ---
+
+[model]
+# name = ""                    # Model ID; empty = provider default (gpt-5 / claude-sonnet-4)
+# temperature = 0.0            # 0 = provider default
+# reasoning_effort = ""        # "low", "medium", "high", "max"
+# context_window = 0           # 0 = auto-detect from provider capabilities
+# max_tokens = 0               # 0 = provider default
+
+# --- Named Models (turnstone, node, eval) ---
+# Define model aliases with per-model overrides. Useful for local model
+# servers or mixing providers. Reference by name with --model flag.
+#
+# [models.local]
+# name = "llama-3-70b"
+# provider = "openai"
+# base_url = "http://localhost:8000/v1"
+# context_window = 8192
+#
+# [models.local.capabilities]
+# supports_vision = false
+# supports_web_search = false
+#
+# [models.claude]
+# name = "claude-opus-4-6"
+# provider = "anthropic"
+
+# --- Database (turnstone, node, console) ---
+
+[database]
+# url = ""                     # postgres://user:pass@host/db or /path/to.db
+                               # env: TURNSTONE_DB_URL
+# SSL params (passed through to SQLAlchemy connection):
+# sslmode = "prefer"           # disable, allow, prefer, require, verify-ca, verify-full
+# sslrootcert = ""             # path to CA cert for verify-ca/verify-full
+# sslcert = ""                 # path to client cert (mTLS)
+# sslkey = ""                  # path to client key (mTLS)
+
+# --- Redis (bridge, console, channel) ---
+
+[redis]
+# url = ""                     # redis://host:6379/0 or rediss://host:6380/0
+                               # env: TURNSTONE_REDIS_URL
+# TLS params (passed through to Redis connection):
+# tls = false                  # enable TLS (also auto-enabled by rediss:// scheme)
+# tls_ca = ""                  # path to CA cert
+# tls_cert = ""                # path to client cert (mTLS)
+# tls_key = ""                 # path to client key (mTLS)
+
+# --- Auth (node, console) ---
+
+[auth]
+# enabled = true               # env: TURNSTONE_AUTH_ENABLED
+# jwt_secret = ""              # HS256 signing secret (min 32 bytes recommended)
+                               # env: TURNSTONE_JWT_SECRET
+# token = ""                   # Static config token for full access
+                               # env: TURNSTONE_AUTH_TOKEN
+
+# --- Logging (turnstone, node, console) ---
+
+[log]
+# level = ""                   # "debug", "info", "warn", "error"
+                               # empty = binary default (warn for CLI, info for servers)
+                               # env: TURNSTONE_LOG_LEVEL
+# json = false                 # JSON output; auto-enabled when stderr is not a TTY
+
+# --- Session (turnstone, node) ---
+
+[session]
+# instructions = ""            # Default system message
+# compact_max_tokens = 32768   # Max tokens for context compaction summary
+# auto_compact_pct = 0.8       # Trigger compaction at this % of context window
+
+# --- Tools (turnstone, node) ---
+
+[tools]
+# timeout = 120                # Tool execution timeout in seconds
+# skip_permissions = false     # Auto-approve all tool calls
+
+# --- Judge (turnstone, node) ---
+
+[judge]
+# enabled = true               # Enable intent validation
+# confidence_threshold = 0.7   # Minimum confidence for heuristic verdicts
+# output_guard = true          # Scan tool output for security signals
+# redact_secrets = true        # Redact detected credentials in output
+
+# --- Memory (turnstone, node) ---
+
+[memory]
+# relevance_k = 5              # Top-K memories for context injection
+# fetch_limit = 50             # Max memories to fetch for ranking
+# max_content = 32768          # Max memory content size in chars
+# nudge_cooldown = 300         # Min seconds between metacognitive nudges
+# nudges = true                # Enable memory nudges
+
+# --- MCP (turnstone, node) ---
+
+[mcp]
+# config_path = ""             # Path to MCP servers config file (JSON)
+# refresh_interval = 14400     # Refresh interval in seconds (default: 4h)
+
+# --- Server (node, console) ---
+
+[server]
+# max_workstreams = 50         # Maximum concurrent workstreams per node
+                               # env: TURNSTONE_MAX_WORKSTREAMS

turnstone/__init__.py

@@ -1,3 +1,3 @@
 """turnstone - Multi-node AI orchestration platform with tool use, agent routing, and cluster simulation."""
 
-__version__ = "0.8.2"
+__version__ = "0.8.9"

turnstone/admin.py

@@ -142,6 +142,189 @@ def _cmd_revoke_token(args: argparse.Namespace) -> None:
         sys.exit(1)
 
 
+# ---------------------------------------------------------------------------
+# TLS commands
+# ---------------------------------------------------------------------------
+
+
+def _cmd_tls_bootstrap(args: argparse.Namespace) -> None:
+    """Initialize CA and issue certs offline."""
+    try:
+        from lacme import CertificateAuthority, FileStore
+    except ImportError:
+        print("lacme not installed. Run: pip install turnstone[tls]", file=sys.stderr)
+        sys.exit(1)
+
+    import contextlib
+    import os
+    from pathlib import Path
+
+    out_dir = Path(args.out)
+    out_dir.mkdir(parents=True, exist_ok=True)
+    with contextlib.suppress(PermissionError):
+        os.chmod(out_dir, 0o700)  # Restrict access — contains CA private key
+
+    store = FileStore(str(out_dir))
+    ca = CertificateAuthority(store, name="turnstone")
+    ca.init(cn="Turnstone CA", validity_days=3650)
+    print(f"CA initialized in {out_dir} (permissions: 0700)")
+
+    # Write CA cert to a well-known location
+    ca_cert_path = out_dir / "ca.pem"
+    ca_cert_path.write_bytes(ca.root_cert_pem)
+    with contextlib.suppress(PermissionError):
+        os.chmod(ca_cert_path, 0o644)
+    print(f"CA cert: {ca_cert_path}")
+
+    # Issue certs for requested domains
+    for domain in args.issue:
+        bundle = ca.issue([domain], validity_hours=48)
+        store.save_cert(bundle)
+        cert_dir = out_dir / "certs" / domain
+        print(f"Issued: {domain} -> {cert_dir}")
+
+    print(f"\nBootstrap complete. {len(args.issue)} cert(s) issued.")
+    print(f"CA and certs written to: {out_dir}")
+
+
+def _cmd_tls_issue(args: argparse.Namespace) -> None:
+    """Request a cert from the console's ACME endpoint."""
+    try:
+        from lacme import SyncClient
+    except ImportError:
+        print("lacme not installed. Run: pip install turnstone[tls]", file=sys.stderr)
+        sys.exit(1)
+
+    import os
+    from pathlib import Path
+
+    console_url = args.console_url
+    if not console_url:
+        console_url = _discover_console_url()
+
+    domains = [args.domain] + args.san
+    directory_url = f"{console_url}/acme/directory"
+    print(f"Requesting cert for {domains} from {directory_url}")
+
+    client = SyncClient(
+        directory_url=directory_url,
+        allow_insecure=True,
+    )
+    bundle = client.issue(domains)
+
+    out_dir = Path(args.out)
+    out_dir.mkdir(parents=True, exist_ok=True)
+    (out_dir / "cert.pem").write_bytes(bundle.cert_pem)
+    (out_dir / "fullchain.pem").write_bytes(bundle.fullchain_pem)
+    (out_dir / "key.pem").write_bytes(bundle.key_pem)
+    os.chmod(out_dir / "key.pem", 0o600)
+
+    print(f"Certificate written to {out_dir}/")
+    print("  cert.pem      (leaf certificate)")
+    print("  fullchain.pem (cert + chain)")
+    print("  key.pem       (private key, 0600)")
+
+
+def _cmd_tls_ca_cert(args: argparse.Namespace) -> None:
+    """Download the CA root certificate from the console."""
+    import httpx
+
+    console_url = args.console_url
+    if not console_url:
+        console_url = _discover_console_url()
+
+    # Use plain HTTP for bootstrap (node may not have CA cert yet)
+    # WARNING: This is trust-on-first-use (TOFU) — verify the fingerprint
+    base = console_url.replace("https://", "http://")
+    url = f"{base}/acme/ca.pem"
+    print(f"Fetching CA cert from {url}")
+    print("WARNING: Fetching over plain HTTP — verify the fingerprint below")
+
+    resp = httpx.get(url)
+    resp.raise_for_status()
+
+    # Show fingerprint for out-of-band verification
+    import hashlib
+
+    fingerprint = hashlib.sha256(resp.content).hexdigest()
+    print(f"CA cert SHA-256: {fingerprint}")
+
+    from pathlib import Path
+
+    Path(args.out).write_bytes(resp.content)
+    print(f"CA cert written to {args.out}")
+
+
+def _cmd_tls_list(args: argparse.Namespace) -> None:
+    """List certificates from the console."""
+    import httpx
+
+    console_url = args.console_url
+    if not console_url:
+        console_url = _discover_console_url()
+
+    url = f"{console_url}/v1/api/admin/tls/certs"
+    headers = {}
+    token = getattr(args, "auth_token", "") or _get_config_token()
+    if token:
+        headers["Authorization"] = f"Bearer {token}"
+    resp = httpx.get(url, headers=headers)
+    resp.raise_for_status()
+    data = resp.json()
+
+    certs = data.get("certs", [])
+    if not certs:
+        print("No certificates issued.")
+        return
+
+    print(f"{'DOMAIN':<30s} {'ISSUED':<22s} {'EXPIRES':<22s}")
+    print("-" * 74)
+    for c in certs:
+        print(f"{c['domain']:<30s} {c['issued_at']:<22s} {c['expires_at']:<22s}")
+
+
+def _get_config_token() -> str:
+    """Try to load auth token from config.toml or environment."""
+    token = os.environ.get("TURNSTONE_AUTH_TOKEN", "")
+    if token:
+        return token
+    try:
+        from turnstone.core.config import load_config
+
+        cfg = load_config("auth")
+        return str(cfg.get("token", ""))
+    except Exception:
+        return ""
+
+
+def _discover_console_url() -> str:
+    """Discover console URL from the services table."""
+    from turnstone.core.storage import get_storage
+
+    try:
+        storage = get_storage()
+    except Exception:
+        print(
+            "No storage configured. Use --console-url or run from a "
+            "directory with a turnstone database.",
+            file=sys.stderr,
+        )
+        sys.exit(1)
+    consoles = storage.list_services("console", max_age_seconds=3600)
+    if not consoles:
+        print(
+            "No console found in services table. Use --console-url explicitly.",
+            file=sys.stderr,
+        )
+        sys.exit(1)
+    return consoles[0]["url"]
+
+
+# ---------------------------------------------------------------------------
+# Entry point
+# ---------------------------------------------------------------------------
+
+
 def main() -> None:
     """Entry point for turnstone-admin CLI."""
     parser = argparse.ArgumentParser(
@@ -171,6 +354,35 @@ def main() -> None:
     p_rt = sub.add_parser("revoke-token", help="Revoke an API token")
     p_rt.add_argument("--token-id", required=True, help="Token ID to revoke")
 
+    # TLS subcommands
+    p_bootstrap = sub.add_parser(
+        "tls-bootstrap",
+        help="Initialize CA and issue certs offline (no running console needed)",
+    )
+    p_bootstrap.add_argument("--out", required=True, help="Output directory for PEM files")
+    p_bootstrap.add_argument(
+        "--issue",
+        action="append",
+        default=[],
+        help="Domain to issue cert for (repeatable)",
+    )
+
+    p_issue = sub.add_parser("tls-issue", help="Request cert from console ACME")
+    p_issue.add_argument("domain", help="Primary domain for the certificate")
+    p_issue.add_argument("--san", action="append", default=[], help="Additional SAN (repeatable)")
+    p_issue.add_argument("--out", default=".", help="Output directory for PEM files")
+    p_issue.add_argument(
+        "--console-url", default="", help="Console URL (discovered from DB if empty)"
+    )
+
+    p_cacert = sub.add_parser("tls-ca-cert", help="Download CA root certificate")
+    p_cacert.add_argument("--out", default="ca.pem", help="Output file path")
+    p_cacert.add_argument("--console-url", default="", help="Console URL")
+
+    p_tlslist = sub.add_parser("tls-list", help="List issued certificates")
+    p_tlslist.add_argument("--console-url", default="", help="Console URL")
+    p_tlslist.add_argument("--auth-token", default="", help="Auth token for admin API")
+
     args = parser.parse_args()
     if not args.command:
         parser.print_help()
@@ -182,5 +394,9 @@ def main() -> None:
         "list-users": _cmd_list_users,
         "list-tokens": _cmd_list_tokens,
         "revoke-token": _cmd_revoke_token,
+        "tls-bootstrap": _cmd_tls_bootstrap,
+        "tls-issue": _cmd_tls_issue,
+        "tls-ca-cert": _cmd_tls_ca_cert,
+        "tls-list": _cmd_tls_list,
     }
     dispatch[args.command](args)

turnstone/api/console_schemas.py

@@ -137,6 +137,9 @@ class ConsoleCreateWsRequest(BaseModel):
         default="", description="Optional first message sent after creation"
     )
     skill: str = Field(default="", description="Skill name (replaces default skills)")
+    resume_ws: str = Field(
+        default="", description="Workstream ID to resume (loads previous conversation)"
+    )
 
 
 class ConsoleCreateWsResponse(BaseModel):
@@ -306,7 +309,10 @@ class SkillInfo(BaseModel):
     agent_max_turns: int | None = None
     notify_on_complete: str = "{}"
     enabled: bool = True
+    priority: int = 0
     allowed_tools: str = "[]"
+    license: str = ""
+    compatibility: str = ""
     scan_status: str = ""
     scan_report: str = "{}"
     scan_version: str = ""
@@ -336,7 +342,10 @@ class CreateSkillRequest(BaseModel):
     agent_max_turns: int | None = None
     notify_on_complete: str = "{}"
     enabled: bool = True
+    priority: int = 0
     allowed_tools: str = "[]"
+    license: str = ""
+    compatibility: str = ""
 
 
 class UpdateSkillRequest(BaseModel):
@@ -359,7 +368,10 @@ class UpdateSkillRequest(BaseModel):
     agent_max_turns: int | None = None
     notify_on_complete: str | None = None
     enabled: bool | None = None
+    priority: int | None = None
     allowed_tools: str | None = None
+    license: str | None = None
+    compatibility: str | None = None
 
 
 class ListSkillsResponse(BaseModel):

turnstone/api/console_spec.py

@@ -82,6 +82,7 @@ from turnstone.api.schemas import (
     CreateTokenRequest,
     CreateTokenResponse,
     CreateUserRequest,
+    DeleteSettingResponse,
     ErrorResponse,
     ListScheduleRunsResponse,
     ListSchedulesResponse,
@@ -751,7 +752,7 @@ CONSOLE_ENDPOINTS: list[EndpointSpec] = [
         "/v1/api/admin/settings/{key}",
         "DELETE",
         "Reset a setting to its default value",
-        response_model=StatusResponse,
+        response_model=DeleteSettingResponse,
         query_params=[
             QueryParam("node_id", "Node ID for node-scoped settings"),
         ],
@@ -842,6 +843,39 @@ CONSOLE_ENDPOINTS: list[EndpointSpec] = [
         error_codes=[400],
         tags=["Admin"],
     ),
+    # --- Admin: TLS / ACME ---
+    EndpointSpec(
+        "/v1/api/admin/tls/ca",
+        "GET",
+        "CA status: initialization state, CN, cert count, cert inventory",
+        tags=["Admin"],
+    ),
+    EndpointSpec(
+        "/v1/api/admin/tls/ca.pem",
+        "GET",
+        "Download CA root certificate (PEM format)",
+        tags=["Admin"],
+    ),
+    EndpointSpec(
+        "/v1/api/admin/tls/certs",
+        "GET",
+        "List all issued TLS certificates",
+        tags=["Admin"],
+    ),
+    EndpointSpec(
+        "/v1/api/admin/tls/certs/{domain}/renew",
+        "POST",
+        "Force-renew a certificate by domain",
+        error_codes=[404, 500],
+        tags=["Admin"],
+    ),
+    EndpointSpec(
+        "/v1/api/admin/tls/certs/{domain}",
+        "DELETE",
+        "Delete a certificate by domain",
+        error_codes=[404],
+        tags=["Admin"],
+    ),
     # --- Observability ---
     EndpointSpec(
         "/health",
@@ -855,6 +889,7 @@ CONSOLE_ENDPOINTS: list[EndpointSpec] = [
 _ALL_MODELS: list[type[BaseModel]] = [
     ErrorResponse,
     StatusResponse,
+    DeleteSettingResponse,
     AuthLoginRequest,
     AuthLoginResponse,
     AuthSetupRequest,

turnstone/api/schemas.py

@@ -8,6 +8,7 @@ as the single source of truth for the generated OpenAPI spec.
 from __future__ import annotations
 
 from enum import StrEnum
+from typing import Any
 
 from pydantic import BaseModel, Field
 
@@ -34,6 +35,14 @@ class StatusResponse(BaseModel):
     status: str = Field(default="ok", examples=["ok"])
 
 
+class DeleteSettingResponse(BaseModel):
+    """DELETE /v1/api/admin/settings/{key} response."""
+
+    status: str = Field(default="ok", examples=["ok"])
+    key: str = Field(description="Dotted setting key that was reset")
+    default: Any = Field(description="Registry default value the setting reverted to")
+
+
 class AuthLoginRequest(BaseModel):
     """POST /v1/api/auth/login request body.
 

turnstone/api/server_schemas.py

@@ -157,8 +157,10 @@ class McpStatus(BaseModel):
 class HealthResponse(BaseModel):
     status: str = Field(examples=["ok", "degraded"])
     version: str = ""
+    node_id: str = ""
     uptime_seconds: float = 0.0
     model: str = ""
+    max_ws: int = Field(default=10, description="Maximum concurrent workstreams")
     workstreams: WorkstreamCounts = WorkstreamCounts()
     backend: BackendStatus | None = None
     mcp: McpStatus | None = None

turnstone/channels/cli.py

@@ -58,6 +58,11 @@ def main() -> None:
         help="HTTP server port (default: $TURNSTONE_CHANNEL_PORT or 8091)",
     )
 
+    # -- TLS -----------------------------------------------------------------
+    parser.add_argument("--ssl-certfile", default=None, help="SSL certificate file for HTTPS")
+    parser.add_argument("--ssl-keyfile", default=None, help="SSL private key file")
+    parser.add_argument("--ssl-ca-certs", default=None, help="SSL CA certs for client verification")
+
     # -- Auth ----------------------------------------------------------------
     parser.add_argument(
         "--auth-token",
@@ -189,7 +194,8 @@ def main() -> None:
                     advertise_host = socket.gethostname()
                 else:
                     advertise_host = args.http_host
-                advertise_url = f"http://{advertise_host}:{args.http_port}"
+                scheme = "https" if args.ssl_certfile else "http"
+                advertise_url = f"{scheme}://{advertise_host}:{args.http_port}"
             service_url = advertise_url
 
             # Register in service registry
@@ -209,11 +215,25 @@ def main() -> None:
                     except Exception:
                         log.exception("channel.heartbeat_failed")
 
+            # TLS: use cert files if available (from bootstrap or TLSClient)
+            ssl_certfile = getattr(args, "ssl_certfile", None)
+            ssl_keyfile = getattr(args, "ssl_keyfile", None)
+            ssl_ca_certs = getattr(args, "ssl_ca_certs", None)
+            if bool(ssl_certfile) != bool(ssl_keyfile):
+                print(
+                    "Both --ssl-certfile and --ssl-keyfile are required for TLS",
+                    file=sys.stderr,
+                )
+                sys.exit(1)
+
             uv_config = uvicorn.Config(
                 channel_app,
                 host=args.http_host,
                 port=args.http_port,
                 log_level="warning",
+                ssl_certfile=ssl_certfile,
+                ssl_keyfile=ssl_keyfile,
+                ssl_ca_certs=ssl_ca_certs,
             )
             server = uvicorn.Server(uv_config)
 

turnstone/channels/discord/bot.py

@@ -306,7 +306,49 @@ class TurnstoneBot:
             await sm.append(event.text)
 
         elif isinstance(event, ApprovalRequestEvent):
-            if self.config.auto_approve or self._should_auto_approve(event):
+            # Evaluate admin tool policies before auto-approve.
+            _policy_handled = False
+            if self.storage is not None:
+                try:
+                    from turnstone.core.policy import evaluate_tool_policies_batch
+
+                    _tool_names = [
+                        it.get("approval_label", "") or it.get("func_name", "")
+                        for it in event.items
+                        if it.get("needs_approval") and it.get("func_name") and not it.get("error")
+                    ]
+                    _tool_names = [n for n in _tool_names if n]
+                    if _tool_names:
+                        verdicts = await asyncio.to_thread(
+                            evaluate_tool_policies_batch,
+                            self.storage,
+                            _tool_names,
+                        )
+                        if any(v == "deny" for v in verdicts.values()):
+                            denied = [n for n, v in verdicts.items() if v == "deny"]
+                            await self.router.send_approval(
+                                ws_id,
+                                event.correlation_id,
+                                approved=False,
+                                feedback=f"Blocked by tool policy: {', '.join(denied)}",
+                            )
+                            await thread.send(
+                                f"*Tool blocked by admin policy: {', '.join(denied)}*"
+                            )
+                            _policy_handled = True
+                        elif all(verdicts.get(n) == "allow" for n in _tool_names):
+                            await self.router.send_approval(
+                                ws_id,
+                                event.correlation_id,
+                                approved=True,
+                            )
+                            await thread.send("*Tool approved by policy.*")
+                            _policy_handled = True
+                except Exception:
+                    log.debug("Tool policy evaluation failed for ws %s", ws_id, exc_info=True)
+            if not _policy_handled and (
+                self.config.auto_approve or self._should_auto_approve(event)
+            ):
                 await self.router.send_approval(ws_id, event.correlation_id, approved=True)
                 await thread.send("*Tool auto-approved.*")
             else:

turnstone/chat.py

@@ -1,42 +0,0 @@
-"""chat.py — Backward-compatibility shim.
-
-All functionality has been moved to submodules:
-  - turnstone.core.session: ChatSession, SessionUI
-  - turnstone.core.tools: TOOLS, AGENT_TOOLS, TASK_AGENT_TOOLS
-  - turnstone.core.edit: find_occurrences, pick_nearest
-  - turnstone.core.sandbox: validate_math_code, execute_math_sandboxed
-  - turnstone.core.safety: is_command_blocked, sanitize_command
-  - turnstone.core.web: strip_html, check_ssrf
-  - turnstone.core.memory: save_message, structured memory facade, etc.
-  - turnstone.ui.colors: ANSI constants and helpers
-  - turnstone.ui.markdown: MarkdownRenderer
-  - turnstone.ui.spinner: Spinner
-  - turnstone.cli: TerminalUI, main, detect_model
-"""
-
-# Re-export public API for backward compatibility
-from turnstone.cli import detect_model, main  # noqa: F401
-from turnstone.core.session import ChatSession, SessionUI  # noqa: F401
-from turnstone.core.tools import AGENT_TOOLS, TASK_AGENT_TOOLS, TOOLS  # noqa: F401
-from turnstone.core.web import strip_html as _strip_html  # noqa: F401
-from turnstone.ui.colors import (  # noqa: F401
-    BLUE,
-    BOLD,
-    CYAN,
-    DIM,
-    GRAY,
-    GREEN,
-    ITALIC,
-    MAGENTA,
-    RED,
-    RESET,
-    YELLOW,
-    bold,
-    cyan,
-    dim,
-    green,
-    red,
-    yellow,
-)
-from turnstone.ui.markdown import MarkdownRenderer  # noqa: F401
-from turnstone.ui.spinner import Spinner  # noqa: F401

turnstone/cli.py

@@ -14,6 +14,7 @@ import textwrap
 import threading
 from typing import TYPE_CHECKING, Any
 
+from turnstone.core.judge import JudgeConfig
 from turnstone.core.session import ChatSession, SessionUI
 from turnstone.core.workstream import Workstream, WorkstreamManager, WorkstreamState
 from turnstone.ui.colors import (
@@ -134,11 +135,42 @@ class TerminalUI(SessionUI):
         """
         pending = [it for it in items if it.get("needs_approval") and not it.get("error")]
 
+        # Evaluate admin tool policies (deny/allow/ask) before prompting.
+        if pending:
+            try:
+                from turnstone.core.policy import evaluate_tool_policies_batch
+                from turnstone.core.storage._registry import get_storage
+
+                storage = get_storage()
+                if storage is not None:
+                    _policy_names = [
+                        it.get("approval_label", "") or it.get("func_name", "")
+                        for it in pending
+                        if it.get("func_name")
+                    ]
+                    if _policy_names:
+                        verdicts = evaluate_tool_policies_batch(storage, _policy_names)
+                        for it in pending:
+                            policy_name = it.get("approval_label", "") or it.get("func_name", "")
+                            verdict = verdicts.get(policy_name)
+                            if verdict == "deny":
+                                it["denied"] = True
+                                it["error"] = f"Blocked by tool policy ('{policy_name}')"
+                                it["needs_approval"] = False
+                            elif verdict == "allow":
+                                it["needs_approval"] = False
+                        pending = [
+                            it for it in items if it.get("needs_approval") and not it.get("error")
+                        ]
+            except Exception:
+                pass  # Best-effort — no policy enforcement on error
+
         with self._print_lock:
             # Print all headers, previews, and heuristic verdicts
             for item in items:
                 if item.get("error"):
                     sys.stdout.write(f"  {red(item['header'])}\n")
+                    sys.stdout.write(f"  {red(item['error'])}\n")
                 else:
                     sys.stdout.write(f"  {yellow(item['header'])}\n")
                 if item.get("preview"):
@@ -162,7 +194,11 @@ class TerminalUI(SessionUI):
 
             # Per-tool auto-approve check
             if self.auto_approve_tools:
-                pending_names = {it.get("func_name", "") for it in pending if it.get("func_name")}
+                pending_names = {
+                    it.get("approval_label", "") or it.get("func_name", "")
+                    for it in pending
+                    if it.get("func_name")
+                }
                 if pending_names and pending_names.issubset(self.auto_approve_tools):
                     return True, None
 
@@ -195,7 +231,11 @@ class TerminalUI(SessionUI):
                     break
 
             if decision in ("a", "always"):
-                tool_names = {it.get("func_name", "") for it in pending if it.get("func_name")}
+                tool_names = {
+                    it.get("approval_label", "") or it.get("func_name", "")
+                    for it in pending
+                    if it.get("func_name") and not it.get("error")
+                }
                 tool_names.discard("")
                 tool_names.discard("__budget_override__")
                 self.auto_approve_tools.update(tool_names)
@@ -752,10 +792,15 @@ def _handle_cluster_command(cmd_line: str, console_url: str | None, auth_token:
 
 
 def detect_model(client: Any, provider: str = "openai") -> tuple[str, int | None]:
-    """Auto-detect model — delegates to :func:`turnstone.core.model_registry.detect_model`."""
+    """Auto-detect model — delegates to :func:`turnstone.core.model_registry.detect_model`.
+
+    CLI always uses fatal=True, so model is never None.
+    """
     from turnstone.core.model_registry import detect_model as _detect
 
-    return _detect(client, provider=provider)
+    model, ctx = _detect(client, provider=provider)
+    assert model is not None  # fatal=True guarantees non-None or SystemExit
+    return model, ctx
 
 
 # ─── Main ──────────────────────────────────────────────────────────────────
@@ -870,6 +915,12 @@ def main() -> None:
         default=5,
         help="Max tools returned per tool search query (default: 5)",
     )
+    parser.add_argument(
+        "--web-search-backend",
+        default="",
+        metavar="BACKEND",
+        help="Web search backend: '' (auto), 'tavily', 'ddg', or 'mcp:server:tool'",
+    )
     parser.add_argument(
         "--resume",
         default=None,
@@ -959,8 +1010,9 @@ def main() -> None:
         default=0.7,
         help="Confidence threshold for judge (default: 0.7)",
     )
-    from turnstone.core.config import apply_config
+    from turnstone.core.config import add_config_arg, apply_config
 
+    add_config_arg(parser)
     apply_config(
         parser,
         ["api", "model", "session", "tools", "console", "auth", "mcp", "database", "judge"],
@@ -980,7 +1032,7 @@ def main() -> None:
     db_url = getattr(args, "db_url", None) or os.environ.get("TURNSTONE_DB_URL", "")
     db_path = getattr(args, "db_path", None) or os.environ.get("TURNSTONE_DB_PATH", "")
     db_pool_size = int(
-        getattr(args, "db_pool_size", None) or os.environ.get("TURNSTONE_DB_POOL_SIZE", "5")
+        getattr(args, "db_pool_size", None) or os.environ.get("TURNSTONE_DB_POOL_SIZE", "2")
     )
     init_storage(db_backend, path=db_path, url=db_url, pool_size=db_pool_size)
 
@@ -1040,6 +1092,20 @@ def main() -> None:
         storage=_get_storage(),
     )
 
+    # apply_config() merges [judge] config.toml values into args as
+    # judge_base_url, judge_api_key, etc.  Output_guard and redact_secrets
+    # default to True, enabling the heuristic guard even when the LLM judge
+    # is disabled via --no-judge.
+    judge_config = JudgeConfig(
+        enabled=args.judge_enabled,
+        model=args.judge_model,
+        provider=args.judge_provider,
+        base_url=getattr(args, "judge_base_url", ""),
+        api_key=getattr(args, "judge_api_key", ""),
+        confidence_threshold=args.judge_confidence,
+        timeout=args.judge_timeout,
+    )
+
     # ChatSession factory — captures shared config for creating workstreams
     def session_factory(
         ui: SessionUI | None,
@@ -1070,7 +1136,9 @@ def main() -> None:
             tool_search=args.tool_search,
             tool_search_threshold=args.tool_search_threshold,
             tool_search_max_results=args.tool_search_max_results,
+            web_search_backend=args.web_search_backend,
             skill=skill or args.skill or None,
+            judge_config=judge_config,
         )
 
     # Create workstream manager and initial workstream

turnstone/console/collector.py

@@ -20,6 +20,7 @@ from typing import TYPE_CHECKING, Any
 import httpx
 
 if TYPE_CHECKING:
+    from turnstone.core.auth import ServiceTokenManager
     from turnstone.mq.broker import RedisBroker
 
 log = logging.getLogger("turnstone.console.collector")
@@ -53,11 +54,14 @@ class ClusterCollector:
         self,
         broker: RedisBroker,
         prefix: str = "turnstone",
-        poll_interval: float = 10.0,
+        poll_interval: float = 15.0,
         discovery_interval: float = 15.0,
-        max_poll_workers: int = 50,
-        http_timeout: float = 5.0,
+        max_poll_workers: int = 200,
+        http_timeout: float = 30.0,
         auth_token: str = "",
+        token_manager: ServiceTokenManager | None = None,
+        tls_verify: Any = True,
+        tls_cert: tuple[str, str] | None = None,
     ):
         self._broker = broker
         self._prefix = prefix
@@ -65,21 +69,52 @@ class ClusterCollector:
         self._discovery_interval = discovery_interval
         self._max_poll_workers = max_poll_workers
         self._http_timeout = http_timeout
+        self._token_manager = token_manager
+        # Static auth header — only used when no token_manager is present.
+        # When a token_manager exists, auth is injected per-request via
+        # extra_headers in _poll_all_nodes to avoid stale JWT expiry.
+        self._static_auth: dict[str, str] | None = None
+        if auth_token and token_manager is None:
+            self._static_auth = {"Authorization": f"Bearer {auth_token}"}
 
         self._lock = threading.Lock()
         self._nodes: dict[str, NodeSnapshot] = {}
         self._running = False
         self._threads: list[threading.Thread] = []
         self._poll_pool = ThreadPoolExecutor(max_workers=max_poll_workers)
-        headers = {}
-        if auth_token:
-            headers["Authorization"] = f"Bearer {auth_token}"
-        self._http_client = httpx.Client(timeout=http_timeout, headers=headers)
+        self._http_client = httpx.Client(
+            timeout=httpx.Timeout(connect=10, read=http_timeout, write=5, pool=http_timeout),
+            limits=httpx.Limits(
+                max_connections=max_poll_workers + 10,
+                max_keepalive_connections=min(max_poll_workers, 200),
+            ),
+            verify=tls_verify,
+            cert=tls_cert,
+        )
 
         # SSE fan-out to browser clients
         self._listeners: list[queue.Queue[dict[str, Any]]] = []
         self._listeners_lock = threading.Lock()
 
+    def upgrade_tls(self, tls_verify: Any = True, tls_cert: tuple[str, str] | None = None) -> None:
+        """Replace the httpx client with one using mTLS context."""
+        old = self._http_client
+        self._http_client = httpx.Client(
+            timeout=httpx.Timeout(
+                connect=10, read=self._http_timeout, write=5, pool=self._http_timeout
+            ),
+            limits=httpx.Limits(
+                max_connections=self._max_poll_workers + 10,
+                max_keepalive_connections=min(self._max_poll_workers, 200),
+            ),
+            verify=tls_verify,
+            cert=tls_cert,
+        )
+        # Don't close old client — concurrent _fetch_node() threads may still
+        # be using it.  It will be GC'd once all references are released, and
+        # the current client is closed in stop().
+        del old
+
     # -- lifecycle -----------------------------------------------------------
 
     def start(self) -> None:
@@ -212,6 +247,7 @@ class ClusterCollector:
                     self._nodes[nid].server_url = meta.get(
                         "server_url", self._nodes[nid].server_url
                     )
+                    self._nodes[nid].max_ws = meta.get("max_ws", self._nodes[nid].max_ws)
 
             # Remove nodes whose heartbeats expired
             lost = [nid for nid in self._nodes if nid not in active_ids]
@@ -233,8 +269,35 @@ class ClusterCollector:
                 log.exception("Poll loop error")
             time.sleep(self._poll_interval)
 
+    @staticmethod
+    def _node_jitter(node_id: str, window: float) -> float:
+        """Deterministic per-node delay within a sliding window.
+
+        Uses a Mersenne prime (2^31 - 1) to hash the node_id into a
+        stable offset so each node is polled at a different point in
+        the cycle.  The offset is consistent across restarts for the
+        same node_id, giving an even spread without randomness.
+        """
+        h = hash(node_id) & 0x7FFFFFFF  # positive 31-bit
+        return (h % 2147483647) / 2147483647 * window  # M31 = 2^31 - 1
+
     def _poll_all_nodes(self) -> None:
-        """Fetch dashboard data from all known nodes in parallel."""
+        """Fetch dashboard data from all known nodes in parallel.
+
+        Submissions are throttled by the thread pool size to avoid a
+        thundering herd — at most ``max_poll_workers`` concurrent HTTP
+        requests are in flight at any time.  Each worker sleeps a
+        deterministic per-node jitter (derived from its node_id) to
+        spread requests across the first half of the poll interval.
+        """
+        # Snapshot current auth header for this poll cycle.  Per-request
+        # headers avoid mutating shared client state (thread-safe).
+        if self._token_manager is not None:
+            poll_headers: dict[str, str] | None = {
+                "Authorization": f"Bearer {self._token_manager.token}"
+            }
+        else:
+            poll_headers = self._static_auth
         with self._lock:
             targets = [
                 (n.node_id, n.server_url)
@@ -245,27 +308,57 @@ class ClusterCollector:
         if not targets:
             return
 
-        futures = {self._poll_pool.submit(self._fetch_node, nid, url): nid for nid, url in targets}
+        jitter_window = self._poll_interval / 2
+
+        def _jittered_fetch(
+            nid: str, url: str, headers: dict[str, str] | None
+        ) -> tuple[dict[str, Any], dict[str, Any]]:
+            delay = self._node_jitter(nid, jitter_window)
+            if delay > 0.1:
+                time.sleep(delay)
+            return self._fetch_node(nid, url, headers)
+
+        futures = {
+            self._poll_pool.submit(_jittered_fetch, nid, url, poll_headers): nid
+            for nid, url in targets
+        }
         for future in as_completed(futures):
             nid = futures[future]
             try:
                 dashboard, health = future.result()
                 self._apply_poll(nid, dashboard, health)
+            except httpx.HTTPStatusError as exc:
+                if exc.response.status_code in (401, 403):
+                    log.warning(
+                        "Auth failure polling node %s: HTTP %d", nid, exc.response.status_code
+                    )
+                else:
+                    log.debug("Failed to poll node %s: HTTP %d", nid, exc.response.status_code)
+                with self._lock:
+                    if nid in self._nodes:
+                        self._nodes[nid].reachable = False
             except Exception:
-                log.debug("Failed to poll node %s", nid)
+                log.warning("Failed to poll node %s", nid, exc_info=True)
                 with self._lock:
                     if nid in self._nodes:
                         self._nodes[nid].reachable = False
 
-    def _fetch_node(self, node_id: str, server_url: str) -> tuple[dict[str, Any], dict[str, Any]]:
+    def _fetch_node(
+        self,
+        node_id: str,
+        server_url: str,
+        extra_headers: dict[str, str] | None = None,
+    ) -> tuple[dict[str, Any], dict[str, Any]]:
         """Fetch /v1/api/dashboard and /health from a single node."""
         base = server_url.rstrip("/")
-        dash_resp = self._http_client.get(f"{base}/v1/api/dashboard")
+        dash_resp = self._http_client.get(f"{base}/v1/api/dashboard", headers=extra_headers)
+        dash_resp.raise_for_status()
         dash_data: dict[str, Any] = dash_resp.json()
         try:
-            health_resp = self._http_client.get(f"{base}/health")
+            health_resp = self._http_client.get(f"{base}/health", headers=extra_headers)
             health_data: dict[str, Any] = health_resp.json()
         except Exception:
+            log.debug("Failed to fetch health from %s", node_id, exc_info=True)
             health_data = {}
         return dash_data, health_data
 
@@ -373,9 +466,12 @@ class ClusterCollector:
         }
 
     def get_nodes(
-        self, sort_by: str = "activity", limit: int = 100, offset: int = 0
+        self, sort_by: str = "activity", limit: int | None = 100, offset: int = 0
     ) -> tuple[list[dict[str, Any]], int]:
-        """Return sorted, paginated node list with per-node counts."""
+        """Return sorted, paginated node list with per-node counts.
+
+        Pass ``limit=None`` to return all nodes (no pagination).
+        """
         with self._lock:
             items = []
             for node in self._nodes.values():
@@ -423,8 +519,15 @@ class ClusterCollector:
         elif sort_by == "name":
             items.sort(key=lambda n: n["node_id"])
 
+        if limit is None:
+            return items[offset:], total
         return items[offset : offset + limit], total
 
+    def get_all_nodes(self) -> list[dict[str, Any]]:
+        """Return all nodes without pagination (for fan-out operations)."""
+        nodes, _ = self.get_nodes(sort_by="activity", limit=None)
+        return nodes
+
     def get_workstreams(
         self,
         state: str | None = None,

turnstone/console/server.py

@@ -37,7 +37,7 @@ from starlette.staticfiles import StaticFiles
 from turnstone.api.console_spec import build_console_spec
 from turnstone.api.docs import make_docs_handler, make_openapi_handler
 from turnstone.console.collector import ClusterCollector
-from turnstone.core.auth import JWT_AUD_CONSOLE, AuthMiddleware
+from turnstone.core.auth import JWT_AUD_CONSOLE, JWT_AUD_SERVER, AuthMiddleware, create_jwt
 
 if TYPE_CHECKING:
     from collections.abc import AsyncGenerator
@@ -133,15 +133,36 @@ _CONSOLE_PROXY_STYLE = "<style>.dashboard-overlay{top:32px!important}</style>"
 
 _VALID_NODE_ID = re.compile(r"^[a-zA-Z0-9._-]+$")
 
+_PROXY_JWT_EXPIRY_SECONDS = 300  # 5 min — ample for any request round-trip
+
 
 def _proxy_auth_headers(request: Request) -> dict[str, str]:
     """Build auth headers for proxied requests to upstream servers.
 
-    Uses the service proxy token (``JWT_AUD_SERVER``) so the upstream node
-    accepts the request.  The user's console-audience JWT is *not* forwarded
-    — it would be rejected by the server's audience validation.
+    Mints a short-lived JWT carrying the real user's identity and scopes
+    so the upstream server records correct audit attribution and enforces
+    scope narrowing.  Falls back to the ServiceTokenManager when no user
+    context is available.
     """
-    # Prefer the auto-rotating ServiceTokenManager when available
+    auth_result = getattr(getattr(request, "state", None), "auth_result", None)
+    jwt_secret: str = getattr(request.app.state, "jwt_secret", "")
+
+    if auth_result is not None and auth_result.user_id and jwt_secret:
+        token = create_jwt(
+            user_id=auth_result.user_id,
+            scopes=auth_result.scopes,
+            source="console-proxy",
+            secret=jwt_secret,
+            audience=JWT_AUD_SERVER,
+            permissions=auth_result.permissions,
+            expiry_seconds=_PROXY_JWT_EXPIRY_SECONDS,
+        )
+        return {"Authorization": f"Bearer {token}"}
+
+    # Fallback: service identity (no user context).
+    # When auth is disabled on the console, auth_result is None, so all proxied
+    # requests use the full-privilege service identity.  This is safe only when
+    # the upstream server also has auth disabled.
     mgr = getattr(request.app.state, "proxy_token_mgr", None)
     if mgr is not None:
         return dict(mgr.bearer_header)
@@ -168,7 +189,7 @@ def _get_server_url(request: Request, node_id: str) -> str | None:
 
 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)
+    nodes = collector.get_all_nodes()
     best_id = ""
     best_headroom = -1
     for n in nodes:
@@ -252,7 +273,7 @@ async def cluster_snapshot(request: Request) -> JSONResponse:
 
 async def cluster_events_sse(request: Request) -> Response:
     collector: ClusterCollector = request.app.state.collector
-    client_queue: queue.Queue[dict[str, Any]] = queue.Queue(maxsize=500)
+    client_queue: queue.Queue[dict[str, Any]] = queue.Queue(maxsize=2000)
 
     async def event_generator() -> AsyncGenerator[dict[str, str], None]:
         loop = asyncio.get_running_loop()
@@ -371,6 +392,7 @@ async def create_workstream(request: Request) -> JSONResponse:
     raw_model = body.get("model", "")
     raw_initial_message = body.get("initial_message", "")
     raw_skill = body.get("skill", "")
+    raw_resume_ws = body.get("resume_ws", "")
     if not isinstance(raw_node_id, str):
         raw_node_id = "" if raw_node_id is None else None
     if not isinstance(raw_name, str):
@@ -381,15 +403,20 @@ async def create_workstream(request: Request) -> JSONResponse:
         raw_initial_message = "" if raw_initial_message is None else None
     if not isinstance(raw_skill, str):
         raw_skill = "" if raw_skill is None else None
+    if not isinstance(raw_resume_ws, str):
+        raw_resume_ws = "" if raw_resume_ws is None else None
     if (
         raw_node_id is None
         or raw_name is None
         or raw_model is None
         or raw_initial_message is None
         or raw_skill is None
+        or raw_resume_ws is None
     ):
         return JSONResponse(
-            {"error": "node_id, name, model, initial_message, and skill must be strings"},
+            {
+                "error": "node_id, name, model, initial_message, skill, and resume_ws must be strings"
+            },
             status_code=400,
         )
     node_id = raw_node_id
@@ -397,9 +424,13 @@ async def create_workstream(request: Request) -> JSONResponse:
     model = raw_model[:128]
     initial_message = raw_initial_message[:4096]
     skill = raw_skill[:256]
+    resume_ws = raw_resume_ws[:64]
 
     from turnstone.mq.protocol import CreateWorkstreamMessage
 
+    auth = getattr(getattr(request, "state", None), "auth_result", None)
+    uid: str = getattr(auth, "user_id", "") or ""
+
     # General pool — push to shared queue, any bridge picks it up
     if node_id == "pool":
         msg = CreateWorkstreamMessage(
@@ -407,6 +438,8 @@ async def create_workstream(request: Request) -> JSONResponse:
             model=model,
             initial_message=initial_message,
             skill=skill,
+            resume_ws=resume_ws,
+            user_id=uid,
         )
         broker.push_inbound(msg.to_json())
         log.debug("Pool dispatch: correlation_id=%s name=%r", msg.correlation_id, name)
@@ -435,6 +468,8 @@ async def create_workstream(request: Request) -> JSONResponse:
         target_node=node_id,
         initial_message=initial_message,
         skill=skill,
+        resume_ws=resume_ws,
+        user_id=uid,
     )
     broker.push_inbound(msg.to_json(), node_id=node_id)
 
@@ -670,17 +705,47 @@ async def _proxy_sse(
 
 @asynccontextmanager
 async def _lifespan(app: Starlette) -> AsyncGenerator[None, None]:
-    # Create async HTTP client for proxy routes
-    headers: dict[str, str] = {}
-    token = app.state.proxy_auth_token
-    if token:
-        headers["Authorization"] = f"Bearer {token}"
-    app.state.proxy_client = httpx.AsyncClient(timeout=30, headers=headers)
-    # Separate client for SSE streams — longer read timeout, shared connection pool
+    # Create async HTTP clients for proxy routes.  Auth headers are NOT baked
+    # in — _proxy_auth_headers() injects a fresh token per-request so JWTs
+    # auto-rotate via ServiceTokenManager instead of expiring after 1 hour.
+    # Size the pool above the fan-out limit to leave headroom for non-fan-out
+    # proxy traffic (UI proxying, SSE streams, etc.).
+    #
+    # Build a ConfigStore so console settings reads get type validation and
+    # caching instead of raw storage.get_system_setting() calls.
+    storage = getattr(app.state, "auth_storage", None)
+    config_store = None
+    if storage:
+        try:
+            from turnstone.core.config_store import ConfigStore
+
+            config_store = ConfigStore(storage)
+        except Exception:
+            log.warning("Failed to initialise ConfigStore", exc_info=True)
+    app.state.config_store = config_store
+    fan_out = (
+        config_store.get("cluster.node_fan_out_limit") if config_store else _NODE_FAN_OUT_LIMIT
+    )
+    app.state.fan_out_limit = fan_out
+    # Build mTLS context for proxy clients if TLS is enabled
+    _tls_mgr = getattr(app.state, "tls_manager", None)
+    _proxy_ssl = _tls_mgr.get_client_ssl_context() if _tls_mgr and _tls_mgr.ca_initialized else None
+    _proxy_verify: Any = _proxy_ssl if _proxy_ssl else True
+
+    app.state.proxy_client = httpx.AsyncClient(
+        timeout=30,
+        limits=httpx.Limits(
+            max_connections=fan_out + 50,
+            max_keepalive_connections=min(fan_out // 4, 100),
+        ),
+        verify=_proxy_verify,
+    )
     app.state.proxy_sse_client = httpx.AsyncClient(
         timeout=httpx.Timeout(connect=5, read=30, write=5, pool=5),
-        limits=httpx.Limits(keepalive_expiry=30),
-        headers=headers,
+        limits=httpx.Limits(
+            max_connections=1100, max_keepalive_connections=100, keepalive_expiry=30
+        ),
+        verify=_proxy_verify,
     )
     # Start scheduler if configured
     scheduler = getattr(app.state, "scheduler", None)
@@ -711,8 +776,84 @@ async def _lifespan(app: Starlette) -> AsyncGenerator[None, None]:
                     "OIDC JWKS prefetch failed — will retry on first login",
                     exc_info=True,
                 )
+    # Register console in service registry so other services can discover it
+    console_url = getattr(app.state, "console_url", "")
+    _console_heartbeat_task: Any = None
+    if console_url and storage:
+        try:
+            storage.register_service("console", "console", console_url)
+
+            # Periodic heartbeat to keep the registration alive
+            import asyncio
+
+            async def _console_heartbeat() -> None:
+                while True:
+                    await asyncio.sleep(30)
+                    try:
+                        storage.heartbeat_service("console", "console")
+                    except Exception:
+                        log.warning("console.heartbeat_failed", exc_info=True)
+
+            _console_heartbeat_task = asyncio.create_task(_console_heartbeat())
+        except Exception:
+            log.warning("Failed to register console service", exc_info=True)
+
+    # TLS: init CA, issue console certs, start renewal
+    tls_mgr = getattr(app.state, "tls_manager", None)
+    if tls_mgr is not None:
+        import socket
+
+        try:
+            if not tls_mgr.ca_initialized:
+                await tls_mgr.init_ca()
+            hostname = socket.getfqdn()
+            cert_hostnames = [hostname, "localhost", "127.0.0.1"]
+            extra_sans = os.environ.get("TURNSTONE_TLS_SANS", "")
+            if extra_sans:
+                cert_hostnames.extend(s.strip() for s in extra_sans.split(",") if s.strip())
+            await tls_mgr.issue_console_certs(cert_hostnames)
+            await tls_mgr.start_renewal()
+            # Re-create proxy clients with mTLS context now that certs are ready
+            client_ctx = tls_mgr.get_client_ssl_context()
+            if client_ctx:
+                await app.state.proxy_client.aclose()
+                await app.state.proxy_sse_client.aclose()
+                app.state.proxy_client = httpx.AsyncClient(
+                    timeout=30,
+                    limits=httpx.Limits(
+                        max_connections=app.state.fan_out_limit + 50,
+                        max_keepalive_connections=min(app.state.fan_out_limit // 4, 100),
+                    ),
+                    verify=client_ctx,
+                )
+                app.state.proxy_sse_client = httpx.AsyncClient(
+                    timeout=httpx.Timeout(connect=5, read=30, write=5, pool=5),
+                    limits=httpx.Limits(
+                        max_connections=1100,
+                        max_keepalive_connections=100,
+                        keepalive_expiry=30,
+                    ),
+                    verify=client_ctx,
+                )
+                # Upgrade collector httpx client for mTLS node polling
+                app.state.collector.upgrade_tls(tls_verify=client_ctx)
+                log.info("tls.proxy_clients.upgraded")
+        except Exception:
+            log.warning("TLS initialization failed — continuing without TLS", exc_info=True)
+
     yield
     # Shutdown
+    if _console_heartbeat_task is not None:
+        _console_heartbeat_task.cancel()
+    # Deregister console from services table
+    if console_url and storage:
+        try:
+            storage.deregister_service("console", "console")
+        except Exception:
+            log.debug("console.deregister_failed", exc_info=True)
+    tls_mgr = getattr(app.state, "tls_manager", None)
+    if tls_mgr is not None:
+        await tls_mgr.stop_renewal()
     if scheduler is not None:
         scheduler.stop()
     await app.state.proxy_sse_client.aclose()
@@ -1467,10 +1608,10 @@ async def admin_list_watches(request: Request) -> JSONResponse:
     if err:
         return err
     collector: ClusterCollector = request.app.state.collector
-    nodes, _ = collector.get_nodes(limit=500)
+    nodes = collector.get_all_nodes()
     client: httpx.AsyncClient = request.app.state.proxy_client
     headers = _proxy_auth_headers(request)
-    sem = asyncio.Semaphore(_NODE_FAN_OUT_LIMIT)
+    sem = asyncio.Semaphore(_get_fan_out_limit(request))
 
     async def _fetch_node(node: dict[str, Any]) -> list[dict[str, Any]]:
         server_url = (node.get("server_url") or "").rstrip("/")
@@ -1509,9 +1650,14 @@ async def admin_list_watches(request: Request) -> JSONResponse:
 _VALID_WATCH_ID = re.compile(r"^[a-fA-F0-9]+$")
 
 # Max concurrent outbound requests when fanning out to cluster nodes.
-# Sized below the default httpx pool limit (100) to leave headroom for
-# other proxy traffic (UI proxying, SSE streams, etc.).
-_NODE_FAN_OUT_LIMIT = 50
+# Must stay below the httpx pool limit (set in _lifespan) to leave
+# headroom for non-fan-out proxy traffic (UI proxying, SSE streams).
+_NODE_FAN_OUT_LIMIT = 200  # fallback; prefer cluster.node_fan_out_limit from storage
+
+
+def _get_fan_out_limit(request: Request) -> int:
+    """Return the fan-out limit cached at startup on app.state."""
+    return int(getattr(request.app.state, "fan_out_limit", _NODE_FAN_OUT_LIMIT))
 
 
 async def admin_cancel_watch(request: Request) -> Response:
@@ -2134,6 +2280,25 @@ async def admin_delete_policy(request: Request) -> JSONResponse:
 
 _VALID_ACTIVATIONS = {"named", "default", "search"}
 
+# Fields that may be updated on installed (readonly) skills.
+# These are local runtime configuration — not part of the SKILL.md spec —
+# so they don't compromise the fidelity of an externally-sourced skill.
+_SKILL_RUNTIME_CONFIG_FIELDS = frozenset(
+    {
+        "model",
+        "temperature",
+        "reasoning_effort",
+        "max_tokens",
+        "token_budget",
+        "agent_max_turns",
+        "auto_approve",
+        "allowed_tools",
+        "enabled",
+        "notify_on_complete",
+        "priority",
+    }
+)
+
 
 def _parse_skill_session_config(body: dict[str, Any]) -> tuple[dict[str, Any], JSONResponse | None]:
     """Parse and validate session config fields from a skill request body.
@@ -2285,7 +2450,10 @@ def _skill_to_response(r: dict[str, Any], resource_count: int = 0) -> dict[str,
         "agent_max_turns": r.get("agent_max_turns"),
         "notify_on_complete": r.get("notify_on_complete", "{}"),
         "enabled": r.get("enabled", True),
+        "priority": r.get("priority", 0),
         "allowed_tools": r.get("allowed_tools", "[]"),
+        "license": r.get("license", ""),
+        "compatibility": r.get("compatibility", ""),
         "scan_status": r.get("scan_status", ""),
         "scan_report": r.get("scan_report", "{}"),
         "scan_version": r.get("scan_version", ""),
@@ -2370,6 +2538,8 @@ async def admin_create_skill(request: Request) -> JSONResponse:
     org_id = str(body.get("org_id", "")).strip()[:64]
     author = str(body.get("author", "")).strip()[:256]
     version = str(body.get("version", "1.0.0")).strip()[:64]
+    license_val = str(body.get("license", "")).strip()[:128]
+    compatibility = str(body.get("compatibility", "")).strip()[:500]
 
     raw_tags = body.get("tags", [])
     if isinstance(raw_tags, list):
@@ -2395,6 +2565,11 @@ async def admin_create_skill(request: Request) -> JSONResponse:
     if activation == "default":
         is_default = True
 
+    try:
+        priority = max(-1000, min(1000, int(body.get("priority", 0) or 0)))
+    except (ValueError, TypeError):
+        priority = 0
+
     if not name:
         return JSONResponse({"error": "name is required"}, status_code=400)
     if not content:
@@ -2418,8 +2593,11 @@ async def admin_create_skill(request: Request) -> JSONResponse:
         tags=tags_str,
         version=version,
         author=author,
+        skill_license=license_val,
+        compatibility=compatibility,
         activation=activation,
         token_estimate=token_estimate,
+        priority=priority,
         **session_fields,
     )
 
@@ -2456,8 +2634,7 @@ async def admin_update_skill(request: Request) -> JSONResponse:
     existing = storage.get_prompt_template(skill_id)
     if existing is None:
         return JSONResponse({"error": "Skill not found"}, status_code=404)
-    if existing.get("readonly"):
-        return JSONResponse({"error": "MCP-sourced skills are read-only"}, status_code=403)
+    is_readonly = bool(existing.get("readonly"))
 
     body = await read_json_or_400(request)
     if isinstance(body, JSONResponse):
@@ -2497,6 +2674,10 @@ async def admin_update_skill(request: Request) -> JSONResponse:
         updates["author"] = str(body["author"]).strip()[:256]
     if "version" in body:
         updates["version"] = str(body["version"]).strip()[:64]
+    if "license" in body:
+        updates["license"] = str(body["license"]).strip()[:128]
+    if "compatibility" in body:
+        updates["compatibility"] = str(body["compatibility"]).strip()[:500]
     if "tags" in body:
         raw_tags = body["tags"]
         if isinstance(raw_tags, list):
@@ -2508,6 +2689,18 @@ async def admin_update_skill(request: Request) -> JSONResponse:
             except (ValueError, TypeError):
                 tag_str = "[]"
             updates["tags"] = tag_str
+    if "priority" in body:
+        try:
+            updates["priority"] = max(-1000, min(1000, int(body["priority"] or 0)))
+        except (ValueError, TypeError):
+            updates["priority"] = 0
+
+    # Installed (readonly) skills: restrict updates to runtime config only.
+    # Spec/content fields are locked to preserve external-source fidelity.
+    if is_readonly:
+        updates = {k: v for k, v in updates.items() if k in _SKILL_RUNTIME_CONFIG_FIELDS}
+        if not updates:
+            return JSONResponse({"error": "No runtime config fields to update"}, status_code=400)
 
     # Snapshot current state for version history before applying update
     existing_versions = storage.list_skill_versions(skill_id)
@@ -2526,7 +2719,7 @@ async def admin_update_skill(request: Request) -> JSONResponse:
     record_audit(
         storage,
         audit_uid,
-        "skill.update",
+        "skill.update.config" if is_readonly else "skill.update",
         "skill",
         skill_id,
         updates,
@@ -2664,6 +2857,16 @@ async def admin_usage(request: Request) -> JSONResponse:
         group_by=group_by,
     )
 
+    # Resolve user_id hex → username for display when grouped by user
+    if group_by == "user" and breakdown:
+        uid_to_name: dict[str, str] = {}
+        for u in storage.list_users():
+            uid_to_name[u["user_id"]] = u.get("username") or u["user_id"]
+        for row in breakdown:
+            raw_key = row.get("key", "")
+            if raw_key and raw_key in uid_to_name:
+                row["key"] = uid_to_name[raw_key]
+
     return JSONResponse({"summary": summary, "breakdown": breakdown})
 
 
@@ -2708,6 +2911,16 @@ async def admin_audit(request: Request) -> JSONResponse:
         until=until,
     )
 
+    # Resolve user_id hex → username for display
+    if events:
+        uid_to_name: dict[str, str] = {}
+        for u in storage.list_users():
+            uid_to_name[u["user_id"]] = u.get("username") or u["user_id"]
+        for ev in events:
+            raw_uid = ev.get("user_id", "")
+            if raw_uid and raw_uid in uid_to_name:
+                ev["username"] = uid_to_name[raw_uid]
+
     return JSONResponse({"events": events, "total": total})
 
 
@@ -3024,20 +3237,18 @@ async def admin_delete_skill_resource(request: Request) -> JSONResponse:
 
 
 def _get_discovery_url(request: Request) -> str:
-    """Get skills discovery URL from DB settings, config.toml, or default."""
+    """Get skills discovery URL via ConfigStore, config.toml, or default."""
     from turnstone.core.config import load_config
     from turnstone.core.skill_sources import DEFAULT_DISCOVERY_URL
 
-    storage = getattr(request.app.state, "auth_storage", None)
-    if storage:
-        try:
-            row = storage.get_system_setting("skills.discovery_url")
-            if row:
-                val = json.loads(row["value"])
-                if val:
-                    return str(val)
-        except (KeyError, json.JSONDecodeError, TypeError, AttributeError):
-            pass
+    # ConfigStore: validated + cached
+    config_store = getattr(request.app.state, "config_store", None)
+    if config_store:
+        val = config_store.get("skills.discovery_url")
+        if val:
+            return str(val)
+
+    # Fall back to config.toml [skills] section
     skills_cfg = load_config("skills")
     url = skills_cfg.get("discovery_url", "")
     if url:
@@ -3203,6 +3414,8 @@ async def admin_skill_install(request: Request) -> JSONResponse:
                 source_url=pkg_source_url,
                 version=parsed.version,
                 author=parsed.author,
+                skill_license=parsed.license,
+                compatibility=parsed.compatibility,
                 activation="named",
                 token_estimate=token_estimate,
                 allowed_tools=allowed_tools_str,
@@ -3394,31 +3607,40 @@ async def admin_delete_memory(request: Request) -> JSONResponse:
 # ---------------------------------------------------------------------------
 
 
-def _publish_config_change(request: Request, *, key: str, node_id: str, action: str) -> None:
-    """Fan out config-reload to all known server nodes (best-effort).
+async def _publish_config_change(request: Request) -> None:
+    """Fan out config-reload to all known server nodes (best-effort, async).
 
-    Uses the collector's node registry and the existing proxy auth
-    mechanism — no MQ dependency.
+    Uses the collector's node registry, the shared async proxy client,
+    and bounded concurrency via the fan-out semaphore.
     """
-    import contextlib
-
-    import httpx
+    # Reload the console's own ConfigStore so cached values stay fresh
+    # (must happen even when collector is absent — e.g. standalone console)
+    config_store = getattr(request.app.state, "config_store", None)
+    if config_store:
+        config_store.reload()
 
     collector = getattr(request.app.state, "collector", None)
     if not collector:
         return
+    client: httpx.AsyncClient = request.app.state.proxy_client
     headers = _proxy_auth_headers(request)
-    with contextlib.suppress(Exception):
-        nodes = collector.get_nodes()
-        for node in nodes.get("nodes", []):
-            url = node.get("url", "")
-            if url:
-                with contextlib.suppress(Exception):
-                    httpx.post(
-                        f"{url}/v1/api/_internal/config-reload",
-                        headers=headers,
-                        timeout=5.0,
-                    )
+    sem = asyncio.Semaphore(_get_fan_out_limit(request))
+
+    async def _notify(url: str) -> None:
+        async with sem:
+            try:
+                await client.post(
+                    f"{url.rstrip('/')}/v1/api/_internal/config-reload",
+                    headers=headers,
+                    timeout=5.0,
+                )
+            except Exception:
+                log.warning("Config reload failed for %s", url, exc_info=True)
+
+    nodes = collector.get_all_nodes()
+    tasks = [_notify(n["server_url"]) for n in nodes if n.get("server_url")]
+    if tasks:
+        await asyncio.gather(*tasks, return_exceptions=True)
 
 
 async def admin_list_settings(request: Request) -> JSONResponse:
@@ -3574,7 +3796,7 @@ async def admin_update_setting(request: Request) -> JSONResponse:
         ip,
     )
 
-    _publish_config_change(request, key=key, node_id=node_id, action="set")
+    await _publish_config_change(request)
 
     return JSONResponse(
         {
@@ -3609,7 +3831,7 @@ async def admin_delete_setting(request: Request) -> JSONResponse:
 
     key = request.path_params["key"]
     try:
-        validate_key(key)
+        defn = validate_key(key)
     except ValueError:
         return JSONResponse({"error": f"Unknown setting: {key}"}, status_code=400)
 
@@ -3629,9 +3851,9 @@ async def admin_delete_setting(request: Request) -> JSONResponse:
         ip,
     )
 
-    _publish_config_change(request, key=key, node_id=node_id, action="delete")
+    await _publish_config_change(request)
 
-    return JSONResponse({"status": "ok", "key": key})
+    return JSONResponse({"status": "ok", "key": key, "default": defn.default})
 
 
 # ---------------------------------------------------------------------------
@@ -3640,21 +3862,16 @@ async def admin_delete_setting(request: Request) -> JSONResponse:
 
 
 def _get_registry_url(request: Request) -> str:
-    """Get the MCP Registry URL from DB settings, config.toml, or default."""
+    """Get the MCP Registry URL via ConfigStore, config.toml, or default."""
     from turnstone.core.config import load_config
     from turnstone.core.mcp_registry import DEFAULT_REGISTRY_URL
 
-    # Check database settings first
-    storage = getattr(request.app.state, "auth_storage", None)
-    if storage:
-        try:
-            row = storage.get_system_setting("mcp.registry_url")
-            if row:
-                val = json.loads(row["value"])
-                if val:
-                    return str(val)
-        except (KeyError, json.JSONDecodeError, TypeError, AttributeError):
-            pass
+    # ConfigStore: validated + cached
+    config_store = getattr(request.app.state, "config_store", None)
+    if config_store:
+        val = config_store.get("mcp.registry_url")
+        if val:
+            return str(val)
 
     # Fall back to config.toml [mcp] section
     mcp_cfg = load_config("mcp")
@@ -3805,8 +4022,9 @@ async def admin_registry_install(request: Request) -> JSONResponse:
 
     # Check max servers
     current = storage.list_mcp_servers()
-    if len(current) >= _MCP_MAX_SERVERS:
-        return JSONResponse({"error": f"Maximum {_MCP_MAX_SERVERS} servers"}, status_code=400)
+    max_servers = _get_mcp_max_servers(request)
+    if len(current) >= max_servers:
+        return JSONResponse({"error": f"Maximum {max_servers} servers"}, status_code=400)
 
     # Fetch the specific server from the registry
     registry_url = _get_registry_url(request)
@@ -3902,7 +4120,15 @@ async def admin_registry_install(request: Request) -> JSONResponse:
 # ---------------------------------------------------------------------------
 
 _MCP_NAME_RE = re.compile(r"^[a-zA-Z0-9._-]+$")
-_MCP_MAX_SERVERS = 50
+_MCP_MAX_SERVERS = 200  # fallback; prefer cluster.mcp_max_servers from storage
+
+
+def _get_mcp_max_servers(request: Request) -> int:
+    """Read cluster.mcp_max_servers via ConfigStore (validated + cached)."""
+    config_store = getattr(request.app.state, "config_store", None)
+    if config_store:
+        return int(config_store.get("cluster.mcp_max_servers"))
+    return _MCP_MAX_SERVERS
 
 
 def _mask_mcp_secrets(server: dict[str, Any], reveal: bool = False) -> dict[str, Any]:
@@ -3940,10 +4166,10 @@ async def _collect_mcp_status(
 ) -> dict[str, dict[str, dict[str, Any]]]:
     """Query all nodes for MCP status. Returns {node_id: {server_name: status}}."""
     collector: ClusterCollector = request.app.state.collector
-    nodes, _ = collector.get_nodes(sort_by="activity", limit=1000, offset=0)
+    nodes = collector.get_all_nodes()
     client: httpx.AsyncClient = request.app.state.proxy_client
     headers = _proxy_auth_headers(request)
-    sem = asyncio.Semaphore(_NODE_FAN_OUT_LIMIT)
+    sem = asyncio.Semaphore(_get_fan_out_limit(request))
 
     async def _fetch(node: dict[str, Any]) -> tuple[str, dict[str, dict[str, Any]] | None]:
         node_id = node.get("node_id", "")
@@ -4087,9 +4313,10 @@ async def admin_create_mcp_server(request: Request) -> JSONResponse:
 
     # Check max servers
     existing = storage.list_mcp_servers()
-    if len(existing) >= _MCP_MAX_SERVERS:
+    max_servers = _get_mcp_max_servers(request)
+    if len(existing) >= max_servers:
         return JSONResponse(
-            {"error": f"Maximum {_MCP_MAX_SERVERS} servers"},
+            {"error": f"Maximum {max_servers} servers"},
             status_code=400,
         )
 
@@ -4291,10 +4518,10 @@ async def admin_delete_mcp_server(request: Request) -> JSONResponse:
 async def _notify_nodes_mcp_reload(request: Request) -> dict[str, Any]:
     """Tell all nodes to re-read the mcp_servers DB table and reconcile."""
     collector: ClusterCollector = request.app.state.collector
-    nodes, _ = collector.get_nodes(sort_by="activity", limit=1000, offset=0)
+    nodes = collector.get_all_nodes()
     client: httpx.AsyncClient = request.app.state.proxy_client
     headers = _proxy_auth_headers(request)
-    sem = asyncio.Semaphore(_NODE_FAN_OUT_LIMIT)
+    sem = asyncio.Semaphore(_get_fan_out_limit(request))
 
     async def _notify(node: dict[str, Any]) -> tuple[str, Any]:
         node_id = node.get("node_id", "")
@@ -4370,6 +4597,7 @@ async def admin_import_mcp_config(request: Request) -> JSONResponse:
     errors: list[str] = []
     audit_uid, ip = _audit_context(request)
     current_count = len(storage.list_mcp_servers())
+    max_servers = _get_mcp_max_servers(request)
 
     for srv_name, cfg in servers.items():
         srv_name = str(srv_name).strip()[:64]
@@ -4379,7 +4607,7 @@ async def admin_import_mcp_config(request: Request) -> JSONResponse:
         if storage.get_mcp_server_by_name(srv_name):
             skipped.append(srv_name)
             continue
-        if current_count >= _MCP_MAX_SERVERS:
+        if current_count >= max_servers:
             errors.append(f"{srv_name}: max servers reached")
             break
 
@@ -4435,6 +4663,161 @@ async def admin_import_mcp_config(request: Request) -> JSONResponse:
     return JSONResponse({"imported": imported, "skipped": skipped, "errors": errors})
 
 
+# ---------------------------------------------------------------------------
+# TLS endpoints
+# ---------------------------------------------------------------------------
+
+
+async def tls_ca_cert(request: Request) -> Response:
+    """GET /v1/api/admin/tls/ca.pem — Download CA root certificate."""
+    from turnstone.core.auth import require_permission
+
+    err = require_permission(request, "admin.settings")
+    if err:
+        return err
+    mgr = getattr(request.app.state, "tls_manager", None)
+    if mgr is None or not mgr.ca_initialized:
+        return JSONResponse({"error": "TLS not enabled"}, status_code=404)
+    return Response(
+        content=mgr.get_root_cert_pem(),
+        media_type="application/x-pem-file",
+        headers={"Content-Disposition": "attachment; filename=turnstone-ca.pem"},
+    )
+
+
+async def tls_ca_status(request: Request) -> JSONResponse:
+    """GET /v1/api/admin/tls/ca — CA status."""
+    from turnstone.core.auth import require_permission
+
+    err = require_permission(request, "admin.settings")
+    if err:
+        return err
+    mgr = getattr(request.app.state, "tls_manager", None)
+    if mgr is None or not mgr.ca_initialized:
+        return JSONResponse({"enabled": False})
+    from turnstone.console.tls import _CA_CN
+
+    certs = mgr.list_certs()
+    return JSONResponse(
+        {
+            "enabled": True,
+            "ca_cn": _CA_CN,
+            "cert_count": len(certs),
+            "certs": [
+                {
+                    "domain": c.domain,
+                    "issued_at": c.issued_at.isoformat(),
+                    "expires_at": c.expires_at.isoformat(),
+                }
+                for c in certs
+            ],
+        },
+    )
+
+
+async def tls_list_certs(request: Request) -> JSONResponse:
+    """GET /v1/api/admin/tls/certs — List issued certificates."""
+    from turnstone.core.auth import require_permission
+
+    err = require_permission(request, "admin.settings")
+    if err:
+        return err
+    mgr = getattr(request.app.state, "tls_manager", None)
+    if mgr is None or not mgr.ca_initialized:
+        return JSONResponse({"certs": []})
+    certs = mgr.list_certs()
+    return JSONResponse(
+        {
+            "certs": [
+                {
+                    "domain": c.domain,
+                    "domains": list(c.domains),
+                    "issued_at": c.issued_at.isoformat(),
+                    "expires_at": c.expires_at.isoformat(),
+                }
+                for c in certs
+            ],
+        },
+    )
+
+
+async def tls_renew_cert(request: Request) -> JSONResponse:
+    """POST /v1/api/admin/tls/certs/{domain}/renew — Force cert renewal."""
+    from turnstone.core.auth import require_permission
+
+    err = require_permission(request, "admin.settings")
+    if err:
+        return err
+    mgr = getattr(request.app.state, "tls_manager", None)
+    if mgr is None or not mgr.ca_initialized:
+        return JSONResponse({"error": "TLS not enabled"}, status_code=404)
+    domain = request.path_params["domain"]
+    try:
+        bundle = mgr.renew_cert(domain)
+        return JSONResponse(
+            {
+                "domain": bundle.domain,
+                "issued_at": bundle.issued_at.isoformat(),
+                "expires_at": bundle.expires_at.isoformat(),
+            },
+        )
+    except ValueError as e:
+        return JSONResponse({"error": str(e)}, status_code=404)
+    except Exception as e:
+        return JSONResponse({"error": str(e)}, status_code=500)
+
+
+async def tls_delete_cert(request: Request) -> JSONResponse:
+    """DELETE /v1/api/admin/tls/certs/{domain} — Delete a certificate."""
+    from turnstone.core.auth import require_permission
+
+    err = require_permission(request, "admin.settings")
+    if err:
+        return err
+    mgr = getattr(request.app.state, "tls_manager", None)
+    if mgr is None or not mgr.ca_initialized:
+        return JSONResponse({"error": "TLS not enabled"}, status_code=404)
+    domain = request.path_params["domain"]
+    if not mgr.delete_cert(domain):
+        return JSONResponse({"error": f"No cert for {domain}"}, status_code=404)
+    return JSONResponse({"deleted": domain})
+
+
+# ---------------------------------------------------------------------------
+# ConfigStore env seeding
+# ---------------------------------------------------------------------------
+
+
+def _seed_config_from_env(config_store: Any, storage: Any) -> None:
+    """Seed ConfigStore settings from environment variables.
+
+    Checks for ``TURNSTONE_{SECTION}_{KEY}`` env vars and writes them
+    to ConfigStore if they aren't already set. This allows container
+    deployments to configure settings before the admin UI is available.
+
+    Only seeds known settings from the registry to avoid storing garbage.
+    Uses config_store.set() for proper validation, serialization, and
+    cache invalidation.
+    """
+    from turnstone.core.settings_registry import SETTINGS
+
+    for key in SETTINGS:
+        env_name = "TURNSTONE_" + key.replace(".", "_").upper()
+        env_val = os.environ.get(env_name)
+        if env_val is None:
+            continue
+        # Only seed if not already stored (check raw storage to avoid
+        # config_store cache, which may not reflect DB state yet)
+        existing = storage.get_system_setting(key)
+        if existing is not None:
+            continue
+        try:
+            config_store.set(key, env_val, changed_by="env")
+            log.info("config.seeded_from_env: %s from %s", key, env_name)
+        except Exception:
+            log.warning("config.seed_failed: %s from %s", key, env_name, exc_info=True)
+
+
 # ---------------------------------------------------------------------------
 # App factory
 # ---------------------------------------------------------------------------
@@ -4450,6 +4833,8 @@ def create_app(
     proxy_auth_token: str = "",
     proxy_token_mgr: Any = None,
     cors_origins: list[str] | None = None,
+    tls_manager: Any = None,
+    console_url: str = "",
 ) -> Starlette:
     """Build the Starlette ASGI application for the console dashboard."""
     _spec = build_console_spec()
@@ -4672,6 +5057,20 @@ def create_app(
                         admin_rescan_skill,
                         methods=["POST"],
                     ),
+                    # TLS / ACME
+                    Route("/api/admin/tls/ca", tls_ca_status),
+                    Route("/api/admin/tls/ca.pem", tls_ca_cert),
+                    Route("/api/admin/tls/certs", tls_list_certs),
+                    Route(
+                        "/api/admin/tls/certs/{domain}/renew",
+                        tls_renew_cert,
+                        methods=["POST"],
+                    ),
+                    Route(
+                        "/api/admin/tls/certs/{domain}",
+                        tls_delete_cert,
+                        methods=["DELETE"],
+                    ),
                 ],
             ),
             Route("/health", health),
@@ -4697,6 +5096,15 @@ def create_app(
     app.state.auth_storage = auth_storage
     app.state.proxy_auth_token = proxy_auth_token
     app.state.proxy_token_mgr = proxy_token_mgr
+    app.state.console_url = console_url
+    app.state.tls_manager = tls_manager
+
+    # Mount ACME responder whenever a TLS manager is configured.
+    # ACMEResponder (lacme 1.0.2+) serves /ca.pem natively.
+    if tls_manager is not None:
+        from starlette.routing import Mount as RouteMount
+
+        app.routes.insert(0, RouteMount("/acme", app=tls_manager.get_responder()))
 
     from turnstone.core.auth import LoginRateLimiter
 
@@ -4781,8 +5189,9 @@ def main() -> None:
         help="Bearer token for polling turnstone-server nodes (default: $TURNSTONE_AUTH_TOKEN)",
     )
 
-    from turnstone.core.config import apply_config
+    from turnstone.core.config import add_config_arg, apply_config
 
+    add_config_arg(parser)
     apply_config(parser, ["console", "redis", "auth"])
     args = parser.parse_args()
 
@@ -4818,13 +5227,13 @@ def main() -> None:
             audience=JWT_AUD_SERVER,
             expiry_hours=1,
         )
-        collector_token = collector_token_mgr.token
-        log.info("console.collector_jwt_minted")
+        log.info("console.collector_token_manager_created")
 
     collector = ClusterCollector(
         broker=broker,
         poll_interval=args.poll_interval,
-        auth_token=collector_token,
+        auth_token=collector_token if collector_token_mgr is None else "",
+        token_manager=collector_token_mgr,
     )
     collector.start()
 
@@ -4843,7 +5252,15 @@ def main() -> None:
         db_backend = os.environ.get("TURNSTONE_DB_BACKEND", "sqlite")
         db_url = os.environ.get("TURNSTONE_DB_URL", "")
         db_path = os.environ.get("TURNSTONE_DB_PATH", "")
-        auth_storage = init_storage(db_backend, path=db_path, url=db_url)
+        auth_storage = init_storage(
+            db_backend,
+            path=db_path,
+            url=db_url,
+            sslmode=os.environ.get("TURNSTONE_DB_SSLMODE", ""),
+            sslrootcert=os.environ.get("TURNSTONE_DB_SSLROOTCERT", ""),
+            sslcert=os.environ.get("TURNSTONE_DB_SSLCERT", ""),
+            sslkey=os.environ.get("TURNSTONE_DB_SSLKEY", ""),
+        )
     except Exception:
         log.info("Console storage not available — admin API disabled, JWT-only auth")
 
@@ -4862,25 +5279,65 @@ def main() -> None:
             audience=JWT_AUD_SERVER,
             expiry_hours=1,
         )
-        proxy_token = proxy_token_mgr.token
-        log.info("console.proxy_jwt_minted")
+        log.info("console.proxy_token_manager_created")
 
     from turnstone.core.web_helpers import parse_cors_origins
 
     cors_origins = parse_cors_origins()
 
+    # TLS: initialize manager if enabled
+    tls_mgr = None
+    # Console URL for service registration — other services use this to discover the console.
+    # Precedence: TURNSTONE_CONSOLE_URL env > auto-detect from bind address.
+    # In Docker Compose, set TURNSTONE_CONSOLE_URL to the service name (e.g. http://console:8090).
+    import socket as _socket
+
+    _console_url_env = os.environ.get("TURNSTONE_CONSOLE_URL", "")
+    if _console_url_env:
+        console_url = _console_url_env
+    else:
+        _advertise_host = args.host
+        if _advertise_host in ("0.0.0.0", "::", ""):
+            _advertise_host = _socket.getfqdn()
+        console_url = f"http://{_advertise_host}:{args.port}"
+    if auth_storage:
+        try:
+            from turnstone.core.config_store import ConfigStore
+
+            _cs = ConfigStore(auth_storage)
+            # Seed ConfigStore from env vars (TURNSTONE_{SECTION}_{KEY})
+            _seed_config_from_env(_cs, auth_storage)
+            if _cs.get("tls.enabled"):
+                from turnstone.console.tls import TLSManager
+
+                tls_mgr = TLSManager(auth_storage, config_store=_cs)
+                # Init CA before create_app so ACME responder can be mounted
+                import asyncio
+
+                asyncio.run(tls_mgr.init_ca())
+                # Upgrade scheme to https if no explicit URL was provided
+                if not _console_url_env:
+                    console_url = console_url.replace("http://", "https://")
+                log.info("TLS enabled")
+        except ImportError:
+            log.warning("TLS enabled but lacme not installed — pip install turnstone[tls]")
+        except Exception:
+            log.warning("TLS initialization failed", exc_info=True)
+
     app = create_app(
         collector=collector,
         broker=broker,
         auth_config=auth_config,
         jwt_secret=jwt_secret,
         auth_storage=auth_storage,
-        proxy_auth_token=proxy_token,
+        proxy_auth_token=proxy_token if proxy_token_mgr is None else "",
         proxy_token_mgr=proxy_token_mgr,
         cors_origins=cors_origins,
+        tls_manager=tls_mgr,
+        console_url=console_url,
     )
 
-    log.info("Console starting on http://%s:%s", args.host, args.port)
+    log.info("Console starting on %s", console_url)
     if auth_config.enabled:
         log.info("Auth: enabled (%d config token(s))", len(auth_config.tokens))
     print("Press Ctrl+C to stop.")

turnstone/console/static/admin.js

@@ -64,6 +64,7 @@ function showAdmin() {
     audit: "admin.audit",
     memories: "admin.memories",
     settings: "admin.settings",
+    tls: "admin.settings",
     mcp: "admin.mcp",
   };
   if (perms) {
@@ -193,6 +194,7 @@ function switchAdminTab(tab) {
     "audit",
     "memories",
     "settings",
+    "tls",
     "mcp",
   ];
   for (var p = 0; p < panels.length; p++) {
@@ -215,6 +217,7 @@ function switchAdminTab(tab) {
   }
   if (tab === "memories") loadAdminMemories();
   if (tab === "settings") loadSettings();
+  if (tab === "tls") loadTlsCerts();
   if (tab === "mcp") loadAdminMcp();
 
   // Update breadcrumb with active tab label
@@ -2057,10 +2060,12 @@ var _settingsSectionOrder = [
   "session",
   "tools",
   "server",
+  "cluster",
   "mcp",
   "ratelimit",
   "health",
   "judge",
+  "skills",
   "memory",
 ];
 
@@ -2070,15 +2075,191 @@ function _settingsSectionLabel(section) {
     session: "Session",
     tools: "Tools",
     server: "Server",
+    cluster: "Cluster",
     mcp: "MCP",
     ratelimit: "Rate Limiting",
     health: "Health",
     judge: "Judge",
+    skills: "Skills",
     memory: "Memory",
   };
   return labels[section] || section;
 }
 
+// ---------------------------------------------------------------------------
+// TLS tab
+// ---------------------------------------------------------------------------
+
+function loadTlsCerts() {
+  var statusEl = document.getElementById("tls-ca-status");
+  var listEl = document.getElementById("tls-cert-list");
+  if (!statusEl || !listEl) return;
+
+  // Fetch CA status and cert list in parallel
+  Promise.all([
+    authFetch("/v1/api/admin/tls/ca").then(function (r) {
+      if (!r.ok) throw new Error("Failed");
+      return r.json();
+    }),
+    authFetch("/v1/api/admin/tls/certs").then(function (r) {
+      if (!r.ok) return { certs: [] };
+      return r.json();
+    }),
+  ])
+    .then(function (results) {
+      var data = results[0];
+      var certData = results[1];
+      while (statusEl.firstChild) statusEl.removeChild(statusEl.firstChild);
+      while (listEl.firstChild) listEl.removeChild(listEl.firstChild);
+
+      if (!data.enabled) {
+        var msg = document.createElement("div");
+        msg.className = "dashboard-empty";
+        msg.textContent =
+          "TLS is not enabled. Set tls.enabled = true in Settings.";
+        statusEl.appendChild(msg);
+        return;
+      }
+
+      // CA status bar
+      var bar = document.createElement("div");
+      bar.className = "tls-ca-bar";
+      var caLabel = document.createElement("span");
+      caLabel.textContent = "CA: " + data.ca_cn;
+      var countLabel = document.createElement("span");
+      countLabel.textContent = "Certificates: " + data.cert_count;
+      bar.appendChild(caLabel);
+      bar.appendChild(countLabel);
+      statusEl.appendChild(bar);
+
+      var certs = certData.certs || [];
+      if (certs.length === 0) {
+        var empty = document.createElement("div");
+        empty.className = "dashboard-empty";
+        empty.textContent = "No certificates issued yet.";
+        listEl.appendChild(empty);
+        return;
+      }
+
+      // Cert rows
+      certs.forEach(function (c) {
+        var row = document.createElement("div");
+        row.className = "admin-row";
+        row.setAttribute("role", "listitem");
+
+        var colDomain = document.createElement("span");
+        colDomain.className = "admin-col";
+        colDomain.textContent = c.domain;
+
+        var colSans = document.createElement("span");
+        colSans.className = "admin-col";
+        colSans.textContent = (c.domains || [c.domain]).join(", ");
+
+        var colIssued = document.createElement("span");
+        colIssued.className = "admin-col";
+        colIssued.textContent = (c.issued_at || "")
+          .slice(0, 16)
+          .replace("T", " ");
+
+        var colExpires = document.createElement("span");
+        colExpires.className = "admin-col";
+        var expires = new Date(c.expires_at);
+        var isExpired = expires < new Date();
+        colExpires.textContent =
+          (isExpired ? "EXPIRED " : "") +
+          (c.expires_at || "").slice(0, 16).replace("T", " ");
+        if (isExpired) colExpires.style.color = "var(--red)";
+
+        var colActions = document.createElement("span");
+        colActions.className = "admin-col admin-col-actions";
+        var renewBtn = document.createElement("button");
+        renewBtn.className = "admin-btn-action";
+        renewBtn.textContent = "Renew";
+        renewBtn.setAttribute(
+          "aria-label",
+          "Renew certificate for " + c.domain,
+        );
+        renewBtn.onclick = function () {
+          tlsRenewCert(c.domain);
+        };
+        var deleteBtn = document.createElement("button");
+        deleteBtn.className = "admin-btn-danger";
+        deleteBtn.textContent = "Delete";
+        deleteBtn.setAttribute(
+          "aria-label",
+          "Delete certificate for " + c.domain,
+        );
+        deleteBtn.onclick = function () {
+          tlsDeleteCert(c.domain);
+        };
+        colActions.appendChild(renewBtn);
+        colActions.appendChild(deleteBtn);
+
+        row.appendChild(colDomain);
+        row.appendChild(colSans);
+        row.appendChild(colIssued);
+        row.appendChild(colExpires);
+        row.appendChild(colActions);
+        listEl.appendChild(row);
+      });
+    })
+    .catch(function () {
+      while (statusEl.firstChild) statusEl.removeChild(statusEl.firstChild);
+      while (listEl.firstChild) listEl.removeChild(listEl.firstChild);
+      var errMsg = document.createElement("div");
+      errMsg.className = "dashboard-empty";
+      errMsg.textContent = "Failed to load TLS status";
+      statusEl.appendChild(errMsg);
+    });
+}
+
+function tlsRenewCert(domain) {
+  showConfirmModal(
+    "Renew Certificate",
+    "Force renew certificate for \u2018" + domain + "\u2019?",
+    "Renew",
+    function () {
+      authFetch(
+        "/v1/api/admin/tls/certs/" + encodeURIComponent(domain) + "/renew",
+        { method: "POST" },
+      )
+        .then(function (r) {
+          if (!r.ok) throw new Error("Renew failed");
+          showToast("Certificate renewed for " + domain);
+          loadTlsCerts();
+        })
+        .catch(function () {
+          showToast("Failed to renew certificate", "error");
+        });
+    },
+  );
+}
+
+function tlsDeleteCert(domain) {
+  showConfirmModal(
+    "Delete Certificate",
+    "Delete certificate for \u2018" + domain + "\u2019? This cannot be undone.",
+    "Delete",
+    function () {
+      authFetch("/v1/api/admin/tls/certs/" + encodeURIComponent(domain), {
+        method: "DELETE",
+      })
+        .then(function (r) {
+          if (!r.ok) throw new Error("Delete failed");
+          showToast("Certificate deleted for " + domain);
+          loadTlsCerts();
+        })
+        .catch(function () {
+          showToast("Failed to delete certificate", "error");
+        });
+    },
+  );
+}
+
+// ---------------------------------------------------------------------------
+// Settings tab
+// ---------------------------------------------------------------------------
+
 function loadSettings() {
   var el = document.getElementById("admin-settings-content");
   if (!el) return;