- fix(ci): trust QA tooling for frozen targets — What Problem This Solves Frozen-target Release Checks run the selected candidate's QA runtime. Candidate 8f382a202ff1e15833394b481615dcdda99b04d7 predates the replay-stable Anthropic mock tool IDs...
- fix(gateway): preserve OpenAI HTTP response outcomes — What Problem This Solves Fixes an issue where OpenAI-compatible HTTP clients could receive a successful Chat Completions or Responses terminal after the agent had resolved with a preserved...
- fix: load ClawHub skill icons in Control UI — What Problem This Solves Fixes an issue where Control UI users searching or opening ClawHub skills would see missing artwork because production browser policy blocked the registry URLs rendered by...
- feat(security): require acknowledgement for install policy warnings — New behavior An external security.installPolicy command can now return warn, letting an authorized operator review a suspicious plugin or skill install before deciding whether to continue. -...
- fix: preserve accepted user turns when preflight or ACP fails — Refs #86592 AI-assisted: yes (Codex). I reviewed and understand the change. What Problem This Solves An accepted user message could be missing from the durable session transcript when required...
- fix(agents): skip repeated workspace bootstrap on continuation turns — Fixes #109656 AI-assisted: This change was developed with Codex. What Problem This Solves Fixes an issue where users who configured contextInjection: "continuation-skip" still received full workspace...
- fix(backup): diagnose invalid archive inputs — What Problem This Solves Fixes an issue where operators running openclaw backup verify received raw Node filesystem errors for missing or directory inputs, while a non-tar garbage file was...
- fix(logging): hide reflected request secrets in provider errors — Related: #119665 Related: #119537 What Problem This Solves Provider APIs can reflect an outbound credential in an error body, structured error metadata, or request id. Those fields can then reach...
- fix(ui): chat send failures were invisible when the session pane was hidden — What Problem This Solves Fixes an issue where operators sending a chat message from the Control UI would see no outcome at all when the send terminally failed while the owning session pane was not...
- Feature request: fallback model chain for compaction and LCM summaryModel — Problem Compaction (agents.defaults.compaction.model) and LCM (plugins.entries.lossless-claw.config.summaryModel) each accept a single model. If that model is rate-limited or unavailable, compaction...
- fix(cron): suggest canonical conversation targets — What Problem This Solves The Cron form mixed channel account identities into recipient suggestions and discarded the canonical thread identity returned by conversations.list. Topic labels that shared...
- fix(control-ui): keep usable multi-profile providers ready — Closes #122241 What Problem This Solves Operators with multiple saved authentication profiles for one model provider could see the provider card report Credentials rejected even though another...
- fix(android): expand truncated assistant messages — What - Preserve and consume the Gateway-owned __openclaw.truncated fact, while retaining the released \n...(truncated)... suffix as the compatibility contract for older/custom Gateways. - Fetch the...
- fix(doctor): report warnings for keyed agent configs — What Problem This Solves Fixes an issue where operators using the canonical keyed agents.entries roster could miss Doctor auth, model-fallback, sandbox, heartbeat, and tool-policy diagnostics, or...
- fix(gateway): batch worker placement session evidence — What Problem This Solves After #124453 removed repeated target identity discovery, Molty still ran near 148% CPU and timed out WebSocket/RPC handshakes. Its 60-second worker-placement retirement...
- feat(ui): review install policy warnings — New behavior An authenticated administrator can review an install-policy warning in the Control UI and deliberately continue that plugin install. - plugins.install accepts the optional literal...
- improve(control-ui): stage slash command arguments in the composer — Refs #123306 — deliberately not closing it. This is the composer/UI phase of that issue. Two decisions it names stay open and are why this PR cannot close it: - Commands outside LOCAL_COMMANDS still...
- feat(models): repeat provider auth setup — What Problem This Solves Adding a second model provider currently requires leaving the auth flow and repeating a broader onboarding/configuration journey. This makes multi-provider setups...
- chore: restore current-main agent E2E baseline — Related: #114492 Additional instructions MUST: Keep Allow edits from maintainers enabled for this PR so maintainers can help update the branch when needed. What Problem This Solves Resolves a problem...
- feat(discord): support private provider endpoints — This PR targets main directly and is non-stacked. Its four owned non-merge commits are replayed directly on main 86bc5aa72644227c8bb21fa6a73c0713221e175e; GitHub reports the published head...
- [AI-assisted] fix(gateway): gate workspace retain dispatch on workerBuild claim — What Problem This Solves Closes #124346 The node workspace retain coordinator dispatches worker.workspace.retain.v1 to every connected node without checking whether that node is a worker host. Nodes...
- build(deps): bump the actions group across 1 directory with 10 updates — [//]: (dependabot-start) ⚠️ Dependabot is rebasing this PR ⚠️ Rebasing might not happen immediately, so don't worry if this takes some time. Note: if you make any changes to this PR yourself, they...
- fix(gateway): surface blocking startup progress — Related to #78954; this PR intentionally does not close the broader core-ready request. Problem Normal managed Gateway startup waits for plugins, channels, workers, and other sidecars. During that...
- fix(agents): guard JSON.parse in hasDuplicateJsonObjectKeys with try/catch — Problem In src/agents/exec-auto-reviewer.ts, hasDuplicateJsonObjectKeys calls JSON.parse on key strings without error handling. Fix Wrap in try/catch, skip key on failure. Testing Regression test for...
- fix(scripts): typecheck hangs forever when tsgo wedges instead of failing — What Problem This Solves Fixes an issue where a local typecheck (pnpm tsgo:core, pnpm check:changed, scripts/pr prepare-gates) would hang indefinitely instead of failing when the tsgo compiler...
- feat(agents): add description field for dynamic agent discovery — Summary Surfaces configured agent descriptions through agents_list so orchestrator agents can understand sub-agent capabilities at runtime. This was originally requested in #7490. Current main...
- feat(qa): run Discord through Crabline's local provider server — Closes #116419 Stack PR 3 of 3. Depends on: - #118067 — neutral Discord provider endpoint runtime (the direct PR base) - #118008 — canonical Crabline server terminology and exact @openclaw/crabline...
- Control UI loses the open session's incognito indicator while the Gateway is disconnected — Problem While the Gateway is disconnected or restarting, the chat pane loses the open session's identity. The incognito indicator is the most alarming case: a user composing in an incognito thread...
- selfhosted Deepseek via vllm reasoning fix — Closes #121597 What Problem This Solves Fixes an issue where users running self-hosted DeepSeek V4 (e.g. DSpark/Flash checkpoints such as yk_deepseek_v4) through the vllm provider never got reasoning...
- Gateway event loop pegged at 100% CPU: broadcastCronBoundSessionChanges does O(events × agent_databases) synchronous statSync on the main thread — Title Gateway event loop pegged at 100% CPU: broadcastCronBoundSessionChanges does O(events × agent_databases) synchronous statSync on the main thread Body Version: 2026.7.2-beta.5 (ee929db), node...
- OpenClaw repo stats for 2026-08-16 — 386427 stars, 81209 forks, 3103 contributors
- OpenClaw sponsors (93 total) — 93 sponsors supporting OpenClaw
- fix: filter skills.sh route query params (#3450)
- feat(claws): align package layers with schema v1 (#3328) — Adds conventional harness profiles, package-root BOOTSTRAP.md, strict OpenClaw validation, portable path hardening, and an official upstream contract pin.
- OpenClaw — Deploy your own personal AI assistant with OpenClaw on a DigitalOcean Droplet® server, a powerful open-source platform that runs entirely on your infrastructure. OpenClaw connects to the messaging...
- OpenClaw surpasses React to become the most-starred software project on GitHub
- OpenClaw vs Subscription AI Assistants: Real Token Costs, Capability Gaps and the Usage Level Where Self-Hosting Wins — OpenClaw wins on cost only at the two extremes of usage: very light use, where a metered token bill...
- I think I found the first OpenClaw voice workflow normal people might actually keep — OpenClaw voice workflows get interesting when you stop trying to automate pizza orders and start automating boring phone errands like haircut bookings
- I finally saw the first OpenClaw document agent I’d trust at a 30-person company, and it wasn’t because of better RAG — A trustworthy OpenClaw document agent for a 30-person company is not the one with the fanciest RAG stack. It’s the one with pinned versions, staged up
- From Agent Security to Neural-Symbolic AI: Three Technical Pivots — In my previous post about actually using OpenClaw, I ended with an idea: perhaps the OpenNebula and OpenClaw tooling I had built could…Continue reading on Medium »
- Behind Hermes’ Rise to the Top: A Web3 Team’s Path Forward — Author: 0xjacobzhao | https://linktr.ee/0xjacobzhaoContinue reading on Medium »
- OpenClaw vs Hermes Agent: The Cage Match for Your Digital Soul — One wants to be your operating system. The other wants to become you. Let’s pick a winner.Continue reading on Generative AI »
- Maid (AI) — Maid is a cross-platform Flutter app for interfacing with GGUF / llama.cpp models locally, and with Ollama and OpenAI models remotely.