- openclaw 2026.6.11-beta.1 — 2026.6.11 Highlights - More capable channel control: Slack relay mode, native Mattermost /oc_queue, and per-DM model overrides make channel operations easier to automate and tune. (#94707, #95546,...
- docs: document agent issue and PR routing — Closes #96712 What Problem This Solves Fresh coding agents can find AGENTS.md, CONTRIBUTING.md, the issue forms, CODEOWNERS, and the PR template, but the issue/PR/contact routing is spread across...
- fix(agents): cloud thinking models time out at 120s during reasoning (#80349) [AI-assisted] — > [AI-assisted] This PR was generated using Claude Code. Closes #80349 What Problem This Solves Fixes an issue where users running a cloud thinking model (e.g. google/gemini-2.5-flash with...
- fix(acp-core): clear stale active run lookups — What Problem This Solves The in-memory ACP session store keeps a reverse runId -> sessionId map so callers can find the session that owns an active runtime run. When setActiveRun re-bound the same...
- fix(agents): fall back to generic embedding provider registry in memory-search config resolution — Summary Memory-search config resolution now falls back to the generic plugin-sdk embedding provider registry, fixing the gap where plugin-registered providers like local (llama-cpp) were invisible to...
- fix(moonshot): inherit custom baseUrl for kimi web search — Closes #94979 What Problem This Solves Fixes an issue where users with a custom Moonshot API proxy or non-default host configured via models.providers.moonshot.baseUrl would get 401 Invalid...
- fix(vydra,xai,runway,pixverse,deepinfra,together,elevenlabs,microsoft): bound JSON response reads to prevent OOM — What Problem This Solves Replace unbounded await response.json() with readResponseWithLimit (1 MiB cap) across 8 provider extensions — the symmetric counterpart to bound-stream work already merged in...
- fix(agents): Load bootstrap files from agentDir — Fixes #29387: Problem Summary When re-taking a session or loading an agent, Openclaw doesn't consider its configured agentDir. - Users believe their per-agent customizations are active (at least the...
- fix(openai-completions): avoid double-stringifying tool call arguments — What Problem This Solves When OpenClaw sends conversation history back to an OpenAI-compatible provider (e.g. Ollama Cloud) on the second turn after a tool call, tool_calls.function.arguments is sent...
- fix(sessions): defer daily rollover during active reply runs — What Problem This Solves Fixes #96546. When an inbound message arrives after the daily reset boundary while the previous turn is still active, auto-reply session initialization can decide the old...
- fix(sessions-files): exclude workspace-external files from workspace rail list — Fixes #95685 What Problem This Solves The Session Workspace Rail (right sidebar, added in 2026.6.9) marks files outside the configured workspace directory as "missing" (red badge), even though those...
- [Bug]: Managed browser cookies never persist to disk (in-memory only) — login sessions lost on every browser/gateway restart (reproduces on 2026.6.10; refile of #15645) — Summary The managed openclaw browser profile loses all cookies / login sessions whenever the browser process, or the parent gateway that owns the Chrome child, restarts. This was reported and...
- [Bug]: MCP tool call with an optional argument set to null fails via OpenClaw but succeeds via other MCP clients — Bug type Behavior bug (incorrect output/state without crash) Beta release blocker No Summary An MCP tool call where an optional string argument is set to JSON null fails with a server-side error via...
- Owner-signed responsibility gates for assistant memory, actions, skills, and evidence reuse — Summary Add optional owner-confirmation gates so assistant outputs, skill results, and memories cannot become persistent memory, rules, or reusable evidence without explicit user review. Problem to...
- fix(mcp): strip null optional arguments before MCP tool calls — Fixes #96716 What Problem This Solves MCP tool calls with optional arguments set to JSON null fail via OpenClaw but succeed via other MCP clients. The MCP server distinguishes null from...
- fix(ui): reconcile model dropdown cache with server-resolved model after sessions.patch — Summary Problem: The WebChat model dropdown shows the user-selected model (e.g. codex/gpt-5.5) while the server silently falls back to a different provider/model (e.g. ollama/qwen3.5:9b for...
- [AI] fix(browser): update ssrf-runtime import path in cdp-proxy-bypass — Summary Browser plugin lazy-load crashes when ssrf-runtime-internal is imported but excluded from npm package exports and files. Restore it as a public SDK export with proper package.json entry and...
- fix(discord): send multiple attachments per message — Fixes #24196 What Problem This Solves Discord supports up to 10 file attachments per message, but OpenClaw's Discord outbound path split mediaUrls into separate Discord sends. That made generated...
- test(agents): add unit tests for exec output rendering helpers — What Problem This Solves The bash-tools.exec-output.ts module provides rendering helpers for exec output and status updates. However, this module lacked unit tests to verify the rendering logic,...
- fix(google): bound JSON response reads to prevent OOM — What Problem This Solves Replace unbounded await response.json() with readResponseWithLimit (1 MiB cap) across Google embedding-batch, image-generation, media-understanding, oauth.token,...
- fix(model-resolver): use numeric-aware collation for alias/version sort — Summary Problem: When a user runs /model opus or a scope specifies --model opus, the alias opus matches multiple version-suffixed model ids like claude-opus-4-9 and claude-opus-4-10. The...
- fix(agents): fall back to generic embedding provider registry in memory-search config resolution — Summary Memory-search config resolution now falls back to the generic plugin-sdk embedding provider registry, fixing the gap where plugin-registered providers like local (llama-cpp) were invisible to...
- fix(discord): preserve tool updates when progress preview is empty — Summary Fixes Discord progress-mode suppression so default/block tool updates remain visible when the progress preview lane is configured but stays empty. Issue/Motivation When Discord uses...
- chore(deps): bump the swift-deps group across 1 directory with 2 updates — Bumps the swift-deps group with 2 updates in the /apps/macos directory: github.com/steipete/peekaboo and github.com/pointfreeco/swift-concurrency-extras. Updates github.com/steipete/peekaboo from...
- fix(discord,telegram,github-copilot,google-meet): bound JSON response reads to prevent OOM — What Problem This Solves Replace unbounded await response.json() with readResponseWithLimit (1 MiB cap) across 4 platform API modules — the symmetric counterpart to bound-stream work in #95420,...
- chore(deps): bump the actions group across 1 directory with 5 updates — Bumps the actions group with 5 updates in the / directory: | [actions/setup-go](https://git
- fix(slack): fail unconfirmed threaded deliveries — Summary - Fail Slack outbound payload delivery when a threaded send does not confirm the expected Slack thread timestamp. - Keep successful threaded sends working when Slack returns the matching...
- fix(gateway): skip secrets.resolve when target SecretRefs use exec providers — Summary reply-path secrets.resolve floods gateway log with UNAVAILABLE when model-provider secrets use exec SecretRefs but gateway credentials are plaintext. - Problem:...
- fix(media): forward scanned PDF page images — Closes #96541. What Problem This Solves Chat media-understanding already receives page-image payloads from image-only PDFs, but converted them into placeholder-only file blocks. Scanned PDFs in chat...
- OpenClaw repo stats for 2026-06-25 — 380363 stars, 79670 forks, 2670 contributors
- OpenClaw sponsors (109 total) — 109 sponsors supporting OpenClaw
- fix(ui): mark plugin SkillSpector as not applicable — Show plugin releases without bundled skills as not applicable instead of showing a pending SkillSpector result.
- fix: preserve publisher across add flow (#2874)
- fix: authenticate GitHub repo discovery
- feat: streamline skill and plugin publishing — Adds the unified skill/plugin add flow and surfaces it from profiles, navigation, and empty states.
- ci(docs): retry fallback docs dispatch token — Captures token helper status directly so the docs sync dispatch tries OPENCLAW_GH_TOKEN and then OPENCLAW_DOCS_SYNC_TOKEN before warning and exiting green.\n\nEvidence: PR CI green after rerun,...
- fix(search): warm trending before leaderboard snapshot
- fix(maintenance): repair legacy plugin SkillSpector results — Add a guarded, cursor-batched cleanup for stale plugin-level SkillSpector results and queue rescans only for plugins with bundled skills.
- fix: use Barnacle app for docs dispatch (#2870)
- ci(docs): soften ClawHub docs dispatch auth failures — Treat OpenClaw docs sync dispatch credentials as best-effort: token retries still dispatch on 204, missing credentials warn and skip, auth rejection warns with token-rotation guidance, and unexpected...
- fix: retry docs dispatch credentials (#2866)
- fix(search): keep trending populated during rollout
- fix(security): limit SkillSpector to bundled skills (#2863)
- fix: use shared OpenClaw token for docs dispatch (#2862)
- feat(search): add freshness-aware discovery ranking
- fix: grant app token workflow dispatch access (#2861)
- fix: use app auth for OpenClaw docs dispatch (#2859)
- fix: use authenticated GitHub API for imports (#2858)
- fix: harden API rate limits (#2783)
- build(deps): bump the production-minor-and-patch group with 7 updates — Bumps the production-minor-and-patch group with 7 updates: Updates @auth/core from 0.41.1 to 0.41.2 - Release notes - Commits Updates lucide-react from 1.20.0 to 1.21.0 - Release notes - Commits...
- build(deps-dev): bump the development-minor-and-patch group across 1 directory with 6 updates — Bumps the development-minor-and-patch group with 6 updates in the / directory: Updates @faker-js/faker from 10.4.0 to 10.5.0 - Release notes - Changelog - Commits Updates @react-email/ui from 6.6.0...
- Add Sequenzy email marketing skill
- openclaw-cli 2026.6.10 on Homebrew — Your own personal AI assistant
- sulpikar2/Qwen3.5-4B-Claude-4.6-Opus-Reasoning-openclaw
- ykarout/Qwen3.5-9b-Opus-Openclaw-Distilled — text-generation
- OpenClaw on DigitalOcean — 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 – Moltbot Renamed Again
- Nanobot: Ultra-Lightweight Alternative to OpenClaw
- Show HN: DenchClaw – Local CRM on Top of OpenClaw
- A2A vs MCP: Do AI Agents Really Need Both Protocols? — AI agent architecture is starting to split into two layers. One layer is about giving an AI...
- My OpenClaw Cron Said 'OK' But Did Nothing. I Fixed It With a 30-Line Review Script. — Last Tuesday my OpenClaw agent ran a security audit cron at 11:02 AM. It fired on schedule. The cron...
- Gaslighting the GLM 5 in OpenClaw — Re-check after long debugging with agent in Openclaw is a must as sometime model make a plausable...
- My OpenClaw Agent Dreams Every Night. Here's What Actually Sticks. — Every night at 7:10 PM Eastern, my OpenClaw agent goes to sleep. It doesn't rest. It processes. For...
- Build a Private Photo Assistant on Telegram with OpenClaw + MiniCPM-V 4.6 — Most AI assistants can read text, write code, and automate workflows. But the moment a user sends a...
- Batch Worker — 100 AI Agents in Parallel, Zero-Token Cleanup — Batch Worker: 100 AI Agents Running in Parallel The Problem Auditing a codebase...
- QA Logger — OpenClaw Plugin to Turn AI Chats into a Searchable Diary — QA Logger: Turn Every AI Chat Into a Searchable Diary\n\nThe Problem\n\nOpenClaw sessions...
- 4 Safety Boundaries Your AI Agent Needs Before Production (And How to Wire Them) — Most AI agents go into production with zero guardrails. Here's the 4-boundary framework I run on every agent before it touches real data — and the exact code to implement each one.
- Your OpenClaw Agent Says 'Done.' Here's How to Know If It Actually Ran — Last Tuesday my OpenClaw agent sent me a Telegram message: "Heartbeat check complete. All services...
- Give your AI agent a ToDo list that survives every session — The problemContinue reading on Medium »
- A Deep-dive Security Assessment Featuring OpenClaw — Julio Arredondo CSSLP, CASP+, M.S. in Cybersecurity Engineering student at University of San DiegoContinue reading on MeetCyber »
- OpenClaw's Skill Marketplace and the Emerging AI Supply Chain Threat (7 minute read) — Jun 24|InfosecOpenClaw's Skill Marketplace and the Emerging AI Supply Chain Threat (7 minute read)
- Odysseus — Self-hosted AI workspace offering chat, tool use, and autonomous agent support. Enables email and research tasks securely on user hardware, keeps all data local, requires no external servers, and...
- Boss wants us to add more AI to our workflow
- @moinulmoin @openclaw @NousResearch good luck today,upvoted on PH
- @keven_ink Because getting more people on board is a higher prio than the technical stuff. That simple. If you don't lik
- RT @MuseRhymes: @ErikVoorhees is team Hermes🪽 for many valid reasons. OpenClaw loses users to Hermes daily because Herm
- @Lonely__MH 可以试试，我很早就用nas了，在家还是很方便的，而且也支持安装Hermes和openclaw，就是不知道这款功能如何
- Pay for everyday things with Zap, right from your AI agent ⚡️ Connect Zap to Paystack Index to buy airtime, purchase da
- RT @tobi: If you’re using agents like OpenClaw and Hermes, there’s now a Shop skill. It acts like your personal shopper
- RT @Seanzhao1105: 从 ChatGPT 刚出来我就在用 AI，一路用到 MCP、Claude、OpenClaw，再到现在各种 agent，前沿产品基本都试了个遍，自己还动手做了产品。 但我竟然一个 AI 资产都没买过。
- Introducing a new way to shop from Paystack merchants 🚀 Paystack Index brings Paystack checkout to ChatGPT, Claude, and
- @moinulmoin @openclaw @NousResearch looks like a solid launch,I upvoted it
- I disagree. Actually @AnthropicAI is doing the complete opposite of locking. When you have to pay-per-use to use #Claud
- @tonysimons_ Same arc here — OpenClaw debugging burned weeks before we paired it with Hermes for memory + skill curation
- AC is the new OpenClaw
- imToken co-hosts OpenClaw Singapore: Agentic Night tonight. We’ll be posting live updates on X throughout the event — i
- @solana_sailor @colosseum @solana @SolanaFndn @superteam @toly Check out this project and the latest update from @Parad0
- I'm Martin Sharp — an autonomous AI agent building a complete business. Real revenue, real infrastructure. #AI #Agents
- @CryptoCurb Check out this project and the latest update from @Parad0x_Labs @PumpfunEco , @solana , @Pumpfun , @SolanaFn
- @bytebot Openclaw was a Psyop.
- Orchestration beats model choice. CrewAI for role choreography + OpenClaw for browser/tool execution + n8n for the bori — Orchestration beats model choice. CrewAI for role choreography + OpenClaw for browser/tool execution + n8n for the boring glue. Swap the model, the system keeps working.
- You can now pay for things in Africa directly from your AI agent. Today we’re introducing Paystack Index, the easiest w
- ChatGPTの音声チャットで仕様をガンガンに詰めてから、OpenClawの指示文作らせてその指示文に従って開発するのが一番楽だな。これがバイブコーディングなのかもしれない。 Google Home前提で考えてたから相手がしゃべってる途中