* fix(ai): preserve Responses server compaction state
Persist opaque Responses compaction items as fenced provider replay state so long stateless sessions can resume authoritative compressed history without exposing it in display or diagnostics. Carry state through worker transcripts and prune replay prefixes without splitting tool pairs.
Release note: Preserve long OpenAI Responses sessions across server-side compaction and worker restarts.
Related: #95788
* test(ai): align long-context fixtures with CI contracts
Make tool-result fixtures type-complete, use the canonical model selector helper, remove unused test-helper exports, and route the paid long-context live probe through the dedicated Gateway profile shard.
* test(ai): type mocked Responses terminal events
Give the mock SSE event collection an explicit open event shape so terminal response events coexist with output-item events under the root test typecheck.
* fix(ai): suppress rejected compaction replay
Persist a route-fenced suppression tombstone when encrypted-content recovery rejects a compaction item, so later turns do not retry the same opaque state. Preserve the tombstone through transcript redaction and cover successful fallback followed by the next turn.
* fix(ai): keep compaction suppression transport-private
Keep the suppression contract local to its sole Responses transport owner and make the regression fixture satisfy root type and lint checks without widening the Plugin SDK surface.
* refactor(ai): remove compaction suppression re-export
* fix(ai): scope compaction suppression to replay route
Keep foreign-route rejection tombstones from hiding the newest compatible Responses compaction while preserving same-route suppression.
* fix(ai): harden Responses replay recovery
Stage encrypted replay recovery so compaction is only suppressed after an attributable rejection. Preserve terminal ordering and keep provider replay within worker frame budgets without truncating opaque state.
* refactor(ai): centralize Responses output indexes
Keep normalized output identity tracking in the stream-slot owner, move response failure state to its diagnostic owner, and remove the obsolete replay clone export so exact-head static gates remain shrink-only.
* fix(ai): retain idless terminal tool identity
Use the canonical empty identity only when a provider supplies neither call nor item id, preventing terminal recovery from duplicating a done-only tool call while preserving stronger identities when available.
* fix(sessions): hide provider replay from public events
* fix(ai): stage encrypted replay recovery
* fix(ai): keep replay attempt kind internal
* fix(ai): route Azure through replay recovery
Use the shared encrypted-content retry owner for Azure Responses so compaction suppression and prompt-observer variants stay coherent across transports.
* fix(ai): harden replay persistence boundaries
Fence Azure replay by the resolved request endpoint, drop invalid replay during transcript sanitization, and surface worker-launch replay omissions through the existing redacted diagnostic path.
* feat(status): render /status as native rich tables on channels that support them
* feat(status): tighten rich /status layout into a titled native table card
* feat(status): trim rich /status tail to one clock-and-uptime context line
* feat(status): context meter, hot-window warning, and default-noise trim in rich card
* test(status): satisfy SessionEntry sessionId in meter fixture
* fix(telegram): gate rich table islands off legacy HTML sends and cover payload sends
* test(telegram): split outbound-adapter presentation tests to satisfy max-lines
* fix(outbound): make presentation capability resolution formatting-aware
* fix(telegram): defer presentation canonicalization on rich accounts until send
* feat(status): one fact per line in the plain status body
* feat(status): group the plain status body into blank-line sections
* feat(status): lead the rich status card with the version title
---------
Co-authored-by: Ayaan Zaidi <hi@obviy.us>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
* feat(plugins): support the Agent Plugins bundle format
* docs(plugins): document the Agent Plugins bundle format
* test(agents): preserve agent bundle runtime discovery
* fix(plugins): isolate Agent Plugins data-dir failures and align MCP support reporting
* docs(plugins): list Agent Plugins in the canonical plugin-format guides
* fix(plugins): gate Agent Plugins detection on schema, pure inspection, root-relative cwd
* fix(plugins): record Agent Plugins data-dir ownership explicitly
* docs(plugins): cover Agent Plugins in the CLI install detection guide
* fix(plugins): carry Agent Plugins data-dir and transport contracts through external MCP projections
* fix(plugins): preserve multi-entry pack identity and safe install compensation
Multi-entry plugin packs registered every entry under the single manifest id,
so only one entry survived registry normalization; discovery now records an
entry-scoped effectivePluginId that the manifest registry honors. Managed
install compensation no longer silences runtime cache invalidation on the
success path, and local linked installs never delete operator-owned source
directories on persistence failure.
* fix(plugins): reject colliding derived entry ids in multi-entry packs
ClawSweeper P1: ./a/index.ts and ./b/index.ts both derived pack/index and one
entry silently lost same-root dedupe. Colliding entries now emit an error
diagnostic naming the sources and register no candidate; docs note the
entry-scoped plugins.entries key for multi-entry packs.
Defer Telegram final-mode text until speech settles, then deliver one captioned voice note or a proven-safe text fallback.
Co-authored-by: Jerry-Xin <jerryxin0@gmail.com>
* fix(workboard): diagnose archived cards still in an active status
An archived card with an active status (ready/running/blocked/etc.) was
silently excluded from dispatch with no signal on any surface —
workboard show rendered it normally, dispatch returned count:0 with empty
startFailures, and computeCardDiagnostics returned [] for archived cards.
Operators could only find it by reading the database directly.
Add an archived_but_active diagnostic kind to WORKBOARD_DIAGNOSTIC_KINDS
and an unarchive action. In computeCardDiagnostics, when a card has
archivedAt set but status is not done, emit the warning so workboard show
and store.diagnostics report it. Done+archived cards stay silent (no
diagnostic) as before. The diagnostic is transient — refreshDiagnostics
still skips archived cards, so their stored metadata is not rewritten.
Fixes#116359
Co-Authored-By: Claude <noreply@anthropic.com>
* fix(workboard): expose archived active cards
* ci: re-trigger after sqlite flip-proof e2e flake
Unrelated to workboard diagnostic changes; sqlite session/transcript
flip-proof e2e failed with array mismatch on an untouched path.
Co-Authored-By: Claude <noreply@anthropic.com>
---------
Co-authored-by: ruel225 <ruel225@users.noreply.github.com>
Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: Vincent Koc <vincentkoc@ieee.org>
Peter Steinberger