Tak Hoffman

fix: keep repair reviews on the pinned base and renew post-flight credentials · fix: pin repair review base and refresh credentials before post-flight publication · fix: give repair edit and validation workers a coherent timeout budget · fix: pin repair review bases and renew post-flight credentials (#494)

fix: give repair edit and validation workers a coherent timeout budget · fix: pin repair worker base during validation · fix: pin repair worker base during validation (#485) · docs: require maintainer edits on pull requests

feat(workspace): add safe workspace reset command · fix(twitch): strip internal tool-trace banners from outbound text

docs(config-channels): fix googlechat groups allow -> enabled to match schema

fix(cli): one-shot agent --local runs never export OTel diagnostics

docs: document local text worker pattern

fix: roll rounded durations into seconds

fix(qa-lab): bound suite runtime gateway JSON response reads

fix(discord): guard JSON.parse against malformed API response bodies · fix: show in-progress status for channel runs

fix: show in-progress status for channel runs

fix(mattermost): bound null-body error response reads

fix(mattermost): truncate draft previews on code-point boundaries

fix(auto-reply): truncate user-facing text on UTF-16 boundary

fix(webchat): make model selector width adaptive to prevent long name…

fix(gateway): report omitted chat-history messages in truncation log

docs(docker): document Claude CLI persistence

fix: UI glitch: config is not visible

Fix memory-wiki bridge self-import loop

fix(agents): reject bind specs with extra colon segments

fix(googlechat): sanitize internal tool-trace lines from outbound text (#90684)

Avoid copying process.env in ingress queue state DB opens

fix(cron): retry isolated setup timeouts

Create a self-improvement skill for claw sweeper · Add autopilot-safe close policy for uncalibrated feature-like PRs · feat(repair): add Crabbox PR lease commands · fix(voice-call): preserve live Twilio streams in stale reaper (#90812)

fix(ui): restore provider usage pill in desktop chat composer [AI]

Fix apply workflow sync before state hydration (#287) · Protect queueable issues from stale automation (#248) · fix(diagnostics): keep recovery scheduling out of the stuck-session warning backoff (#92752) · fix(memory): keep memory_search in transient qmd mode (#92639)

fix(anthropic-vertex): stop re-marking cache_control on transport-budgeted payloads (#92387) · fix(anthropic-vertex): stop re-marking cache_control on transport-budgeted payloads

fix: autostash workflow rebases · fix: keep skill toggles keyed by skill identity

fix(ui): require user intent for chat sessions

fix #90452: Regression: Heartbeat exec completion still shows generic fallback text instead of actual output · fix(agents): do not refresh lastUsedAt on MCP lease release

fix(agents): prevent ReDoS in background-session name derivation

Fix Control UI CJK IME composition

[codex] Simplify proof nudge schedule · Simplify proof nudge schedule · fix(minimax): exempt M3 from thinking-disabled wrapper · fix(agents): detect unsigned thinking-only stall when reasoning payload inflates payloadCount

[codex] Fix proof nudge schedule gate · Fix proof nudge schedule gate (#254)

Handle delayed proof nudge schedules · Handle delayed proof nudge schedules (#249) · fix(telegram): isolate verbose status after streamed finals

Gate proof nudges to 5am Central · Run proof nudges at 5am Central · Allow proof nudges to comment on PRs · Keep proof-nudge comments inert

fix(memory-core): keep startup cron retries quiet

Create a self-improvement skill for claw sweeper · Require routable ClawSweeper commands before fast ack

Fix issue-author re-review commands with trailing context · Require routable ClawSweeper commands before fast ack · fix(agents): classify expired thinking signatures

Fix issue-author re-review commands with trailing context · fix(config): preserve empty plugin allowlist · fix(agents): reuse cached subagent registry reads · fix(gateway): cache single session row child indexes

Reduce false-positive priority and merge-risk labels for unrelated CI noise · Duplicate command ack comments can survive current prune race · Add autopilot-safe close policy for uncalibrated feature-like PRs · Suppress empty maintainer-ruling fields in visualize output

Add maintainer override for issue build refusals · feat: add visualize commands · feat: add issue build override · fix: shallow fetch event state checkout

docs: clarify config default review policy · Add catch-all impact and merge-risk labels with required justification · Add review metrics digest near PR report stats · Add AGENTS.md policy status to review details

docs: add ClawSweeper review policy to AGENTS · test: hello world · fix: Refine PR template for review state · Treat AGENTS.md as review policy during ClawSweeper reviews

feat: add OpenClaw PR surface stats · Hide PR egg outside OpenClaw repos · Require AGENTS.md in review prompts · feat: add OpenClaw PR surface stats (#175)

feat(crabbox): add PR desktop lease workflow · fix(ci): route ClawSweeper comment commands · fix(review): shorten freshness line · fix(review): show durable comment freshness

revert(assist): remove visual explainer feature · fix(assist): retry visual explainer sanitizer failures · fix(assist): keep visual dispatch payload within GitHub limit · feat(assist): add visual PR explainer

Contributor / Maintainer Start Here · feat(review): add feature showcase labels · [codex] Add merge command alias · revert(review): restore previous comment formatting

Move motivation section to top of pull request template · fix(hatch): preserve exact records during reconcile · fix(review): explain label transitions · [codex] Remove merge risk prefix

fix(github): preserve clawsweeper proof labels · fix(ci): preserve Barnacle proof labels · fix(ci): authenticate proof verdict markers · fix(ci): honor exact-head proof verdicts

fix(ci): clear Mantis command reactions · docs(mantis): prefer openclaw mantis mention · ci(mantis): allow ClawSweeper telegram proof agent · fix(mantis): remove ambiguous GitHub trigger mention

Fit priority label descriptions to GitHub limit · Add ClawSweeper triage priority labels e.g. P0, P1, P2, P3 · Merge pull request #77 from openclaw/codex/fix-priority-label-descriptions · fix(labels): fit priority descriptions to GitHub limit

fix(active-memory): allow active-memory to support custom recall tools · fix(active-memory): allow active-memory to support custom recall tools (#77906)

fix(active-memory): allow active-memory to support custom recall tools

fix(active-memory): skip sub-agent gracefully when no memory tools registered (#77506)

fix(active-memory): skip scoped topic channelId for embedded recall run in Telegram forum topics (#76704)

fix(active-memory): use bundled recall tool · fix: support explicit active-memory chat types · fix(active-memory): preserve setup time outside recall timeout · feat(active-memory): return partial transcript on timeout

Fail package update on unhealthy gateway restart · Hooks: stabilize session-memory filenames · fix(plugins): share bundled runtime SDK alias prep · docs: centralize harness syntax guidance

Tighten group silent caution prompt · Align silent reply prompt guidance · fix(comfy): read config from plugins.entries instead of models.providers · fix: tweak group silent caution prompt

fix(plugins): share bundled runtime SDK alias prep · Fix(models add): OpenAI Codex model add metadata · OpenAI image generation follow-up · Add packed CLI smoke checks for release packaging

fix: recover gateway config after invalid startup · feat(commands): gate /models add with modelsWrite · feat: Add /models add hot-reload model registration · feat(commands): gate /models add with modelsWrite (#70321)

docs: restore general multi-gateway guidance · docs: front-load rescue bot quickstart · docs: clarify rescue bot gateway setup · feat: add soft reset command

fix(bootstrap): close silent 10% content gap in trim ratios

feat: Add profile-aware rescue installer options · Add profile-aware rescue installer options

feat(workspace): add safe workspace reset command · fix(bootstrap): workspace bootstrap prompt routing · Add profile-aware rescue installer options

Hooks: stabilize session-memory filenames · fix(startup): prioritize bootstrap on fresh sessions · fix(docs): add active memory speed recommendations

feat: add per-agent web_fetch ssrf overrides · fix(context-window): Tighten context limits and bound memory excerpts · fix(context-window): Tighten context limits and bound memory excerpts (#67277)

fix(active-memory): Move active memory recall into the hidden prompt prefix · fix:(trace command): Improve trace raw diagnostics and trace command UX · Clarify failover log source and target · fix(active-memory): Move active memory recall into the hidden prompt prefix (#66144)

docs: centralize harness syntax guidance · Fix dreaming replay, repair polluted artifacts, and gate wiki tabs · fix(agent-init): move session startup context into the runtime · Fix active-memory recall runs when mx-claw is enabled

fix(qa-lab): release smoke · Add proxy capture stack and QA Lab inspector · feat(ui): render assistant directives and add embed tag · feat(qa-lab): Add proxy capture stack and QA Lab inspector (#64895)

fix(cli): route gateway media sends through sendMedia · fix: include memory plugins in gateway startup · fix(skills): add missing opening --- to taskflow and taskflow-inbox-triage SKILL.md frontmatter · feat(ui): render assistant rich output in Control UI

Add Active Memory recall plugin · Feat: Add Active Memory recall plugin (#63286)

feat: Add first-class infer CLI for inference workflows

fix: normalize cached prompt token accounting

agents: remove prompt-time tool-result guard and add prompt-cache lane · CI: make bad-barnacle bypass PR auto-response · ui: add chat thinking selector

docs: trim PR template root-cause boilerplate · fix: honor tools default account · fix: honor allowlist default account · fix: honor approve default account