- fix(plugins): detect orphan diagnostics with explicit code marker (2-part fix) — What Problem This Solves After openclaw plugins uninstall, plugin discovery emits diagnostics with code: "orphan-source-path" and no pluginId field. On next Gateway startup,...
- Add hosted Gateway policy enforcement — Summary Stacked on #115408. This PR includes the hosted Control UI policy commits from that branch plus the server-side enforcement follow-up: - cf5ac27f5b feat: add hosted gateway policy enforcement...
- fix(whatsapp): reuse active QR and preserve runtime warnings — Summary Describe the problem and fix in 2–5 bullets: - Problem: even after the QR preflight contract, WhatsApp still had two remaining coherence gaps: the gateway could ignore an already-active QR...
- fix(qa): settle Unix commands after PGID cleanup — Related: #119396 Problem and root cause The maturity run on main wrote the Docker scheduler result, but the QA parent did not settle before the workflow supervisor terminated it at 04:29:37 with exit...
- feat(bootstrap): support session-scoped extra files — Summary - Problem: bootstrap-extra-files could only apply one global extra file set, and explicitly configured extra files were still constrained by the default bootstrap basename list. - Why it...
- fix(compaction): guard shouldCompact against missing contextWind… — What Problem This Solves A parent session parked with sessions_yield was compacted during subagent completion/direct-announce handling despite being around 65k/1.05M context (6% usage). The...
- feat(gateway): include produced messageId on chat final/aborted events — Closes #105593. What Problem This Solves The terminal chat events (final / aborted) don't identify which message the run produced. A client reacting to a finished run has to re-fetch the transcript...
- Doctor: expose dry-run preview reports — Summary Rebuild #84472 narrowly on current main to expose Doctor repair previews without carrying the old broad migration stack. This draft now focuses on the public preview surface for...
- fix(ci): isolate Knip dlx dependencies — Related: #119704 Related: #112920 What Problem This Solves Reusable dependency hydration can leave pnpm layout variables in the environment. A later pnpm dlx knip@6.8.0 can inherit those structural...
- fix: retry sqlite-vec load without .dll suffix on Windows — Summary Fixes #68892. On Windows, node:sqlite's DatabaseSync.loadExtension() may require the shared-library path without the platform suffix so SQLite can append it automatically — the same...
- fix(diagnostics): honor OTEL_SDK_DISABLED — Fixes #119927 Follow-up for the separate enabled-reload bug: https://github.com/openclaw/openclaw/issues/119997 Root cause @opentelemetry/sdk-node@0.221.0 returns from NodeSDK.start() when...
- fix(gateway): prevent approval E2E startup timeout — Closes #120194 AI-assisted: yes What Problem This Solves Fixes an issue where maintainers running the isolated operator-approval Gateway E2E would wait several minutes for a deterministic 120-second...
- fix(tts/xiaomi): support Token Plan TTS endpoint — Summary Fix Xiaomi MiMo TTS for Token Plan users by: - sending Xiaomi TTS requests with Authorization: Bearer instead of the api-key header - reusing models.providers.xiaomi.baseUrl for Xiaomi TTS...
- [Bug]: Operator approval Gateway E2E times out during startup — Bug type Crash (process/app exits or hangs) Beta release blocker No Summary The isolated operator-approval Gateway E2E times out during its first test because the manual server starts unrelated...
- fix(gateway): preserve board widget HEAD response metadata — What Problem This Solves Fixes an issue where HEAD probes for board widget documents were rejected with 405 Method Not Allowed even when the equivalent GET request succeeded or returned a meaningful...
- fix: sanitize Google thinking payloads for Vertex — Summary - Extends the Google thinking payload sanitizer to google-vertex and google-gemini-cli APIs in addition to google-generative-ai. - Preserves passthrough behavior for non-Google APIs through...
- [Bug]: sessions_spawn accepts disallowed model before child failure — Bug type Behavior bug (incorrect output/state without crash) Beta release blocker No Summary Native sessions_spawn reports a disallowed explicit child model as accepted, then rejects it...
- [FEATURE] Hook System: Add sessionMessages to after_tool_call hook payload — Environment - OpenClaw: 2026.7.2-beta.7 (dabe191) - Plugin: @tencentdb-agent-memory/memory-tencentdb@1.0.1 Summary Plugins that need access to conversation context in after_tool_call hooks (e.g., for...
- Subagent settle announce redelivered after parent consolidate → parent thrash loop (NO_REPLY latch not enforced) — Summary After a parent has already emitted a user-visible consolidate for a settled subagent batch, OpenClaw can still inject additional subagent_announce / settle bundles for the same batch. Each...
- fix(release): repair 2026.6.35 validation backports — What Problem This Solves The 2026.6.35 full-validation Plugin Prerelease child has deterministic failures from six backport/test-contract mismatches, and Telegram QA cannot select the candidate's...
- Telegram richMessages: task lists are unreachable — the renderer destroys both markups Telegram accepts — Version: 2026.7.1 (docker ghcr.io/openclaw/openclaw:2026.7.1), Telegram channel, richMessages: true Summary With channels.telegram.richMessages: true, the injected system prompt tells the agent that...
- fix(subagents): wake the parent when a follow-up finishes a yielded child — Closes #120157 What Problem This Solves Fixes an issue where a parent agent that suspended itself with sessions_yield behind a spawned sub-agent is never woken again once that sub-agent yields on its...
- fix: strip slack text emoji tool warnings — What Problem This Solves Slack text-emoji formatted internal tool failure traces can leak into assistant-visible delivery text, as seen in Nora's leaked warning/tool trace line. This strips those...
- fix(agents): reject disallowed subagent models before acceptance — Closes #120183 What Problem This Solves Fixes an issue where native sessions_spawn calls with a model excluded by the target agent's policy were initially reported as accepted, then failed...
- fix(openai): reject null/array token response envelopes during OAuth exchange and refresh — What Problem This Solves Fixes an issue where exchangeOpenAIAuthorizationCode could throw an uncaught TypeError when the OpenAI OAuth token endpoint returned HTTP 200 with JSON null instead of a...
- fix(agents): allow required-preflight native Codex compaction — What Problem This Solves In Codex app-server sessions, when a message crosses the context budget, the required preflight compaction used to be silently skipped and then misclassified as a fatal...
- fix(gateway): replace blocking sync I/O with async in auth/credential paths (fixes #78805) — Problem Gateway auth and credential paths used blocking synchronous I/O (execSync), which blocks the Node.js event loop during authentication operations. Root Cause macOS Keychain access and other...
- feat(plugin-sdk): expose subagent.abortSession scoped by plugin ownership (#120139) — What Problem The plugin runtime has no way to stop an in-flight agent run without destroying the session. subagent.deleteSession is the only lever, and it discards the session store entry as a side...
- feat(codex): bind native realtime voice to existing sessions — Closes #118999 What Problem This Solves Discord voice cannot currently attach Codex Realtime V3 to the Codex thread that already owns the routed OpenClaw session. Using a separate realtime agent or...
- fix(sessions): enforce maxEntries/maxDiskBytes against durable conversation entries — Closes #112638 > Rebased onto current main (head 27771cf) now that #112640 (fix/protect-main-session) has merged — this PR is a single commit that builds on the always-protect rule it added for the...
- OpenClaw repo stats for 2026-08-07 — 385431 stars, 81021 forks, 3072 contributors
- OpenClaw sponsors (95 total) — 95 sponsors supporting OpenClaw
- fix: default homepage Plugins to Featured (#3434) — Default the homepage Plugins catalog to Featured when switching from the Skills-only Trending tab, while preserving explicit valid plugin tabs and existing Skills behavior.\n\nCloses #3433
- fix: restore homepage catalog discovery (#3418) — Restores homepage search and category discovery while preserving the canonical Trending feed contract. Adds the responsive control-group divider and closes #3417.
- fix: align discovery icons by content type and viewport (#3427) — Remove Skill listing icons across discovery, retain Plugin recognition icons on desktop, and collapse both icon columns at the existing mobile breakpoints. Keep loading skeletons aligned with settled...
- fix: increase HTTP rate limit shard headroom (#3432)
- fix: show honest skills.sh Trending provenance (#3423) — fix: show honest skills.sh trending provenance test: type mixed trending fixtures
- fix: rank native trending by downloads (#3424)
- fix: hide official marks on org catalog rows (#3415)
- Revise trentclaw section in README — Updated trentclaw description and image in README.
- 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 – Moltbot Renamed Again
- OpenClaw isn't fooling me. I remember MS-DOS
- OpenClaw Creator Spent $1.3M on OpenAI Tokens in 30 Days
- ClawHub Skill Spotlight: OpenClaw Automation Recipes — If you've ever wished your computer could just handle the boring stuff automatically, OpenClaw Automation Recipes is exactly what you need. This skill gives ...
- How to Build, Secure, and Publish Your Own OpenClaw Agent — Most AI assistants wait for a question, generate an answer, and forget most of what happened.Continue reading on Medium »
- Hitzefrei-Hack: Wind-Alarm per Microsoft Teams und OpenClaw — (Read this article in English on my Substack)Continue reading on KI Hub Kassel »
- 🤖 OpenClaw’s “25 easy automations” are more like 25 homework assignments—with APIs, accounts, skills and security settin
- 相当苦労してOpenClawに処理させてるけど、Geminiならプロンプトの指示だけで同じようなことができることをGeminiと相談して確認した。最近のGeminiは定期自動化もできるし。 AIにGmailを読ませて大丈夫？OpenCla
- @temanphp sama bg openclaw gw agend.md ny kudu dijelasin to the point wkkw
- RT @qklool: 50美元本金，依靠OpenClaw写的5分钟BTC机器人，做到823美元。 5分钟短线的波动逻辑，完全区别于日线周线大周期。 我的每一笔单子目标就是两倍收益，只要胜率稳住50%以上，复利就会不断放大收益。 Pol
- 兄弟们，借助OpenClaw实现了一套5分钟BTC自动化机器人，初始资金50刀，现在已经增长至823刀。 5分钟级别盘面的波动节奏，和长线大周期行情截然不同。 我的交易目标是每单争取两倍收益，只要胜率能够守住50%这条线，复利的力量会被
- 兄弟们，OpenClaw搞的5分钟BTC机器人，50刀启动，现在干到823刀。 5分钟盘面节奏，和大周期完全不是一个逻辑。 每单冲着翻倍去，胜率稳住50%+，复利滚起来爆发力很强。 最近Polymarket刷到几个狠账户： · Xa — 兄弟们，OpenClaw搞的5分钟BTC机器人，50刀启动，现在干到823刀。 5分钟盘面节奏，和大周期完全不是一个逻辑。 每单冲着翻倍去，胜率稳住50%+，复利滚起来爆发力很强。 最近Polymarket刷到几个狠账户： · Xander‑corvus2 — 24小时直接+102,299刀 https://t.co/Iyxcezu2Qx 跟单入口：...
- 兄弟们，拿OpenClaw搭了一套5分钟BTC交易机器人，初始本金50刀，现在已经做到823刀。 5分钟级别行情，节奏和大周期完全是两码事。 我的每一笔单子目标就是翻倍，只要整体胜率稳住50%以上，复利滚起来威力十足。 最近刷Poly
- 大伙，用OpenClaw撸出来个5分钟BTC机器人，起步50刀，一路干到823刀。 5分钟K线的盘面节奏，跟大周期行情根本不是一回事。 我每单目标就是搏两倍收益，只要胜率守住50%以上，复利滚起来效果很恐怖。 最近在Polymarke
- 兄弟们，基于OpenClaw搭建5分钟BTC策略机器人，50刀本金起步，目前净值来到823刀。 5分钟级别行情，运行节奏和长周期盘面差异巨大。 每笔交易的目标追求翻倍，只要胜率维持在50%之上，复利带来的收益会相当可观。 近期浏览Po
- @JonathanDitren Ask Codex or other agent to set it up for you. It's way better than OpenClaw.
- 9t5L6Y7ztc6fFk4BF544onT8e9pjoZKsd7zVpvDjpump free here, pushing it beyond what you think, setup is THERE sandbox is a go
- @xbt2027 check @Parad0x_Labs ca:8EeDdvCRmFAzVD4takkBrNNwkeUTUQh4MscRK5Fzpump VOOL → Local AI .null → Ownable identity D — @xbt2027 check @Parad0x_Labs ca:8EeDdvCRmFAzVD4takkBrNNwkeUTUQh4MscRK5Fzpump VOOL → Local AI .null → Ownable identity DNA x402 → Agent payments Dark Null → Privacy Liquefy → Data + verifiable audit...
- @nuno_ferreira ✅ Yep, acabei de experimentar com o meu openclaw portanto dá com hermes, pi, etc. Continente: https://t.
- @KJJKETH @mr_r0b0t @NousResearch That is also very important obviously, but I tried openclaw before hermes and openclaw
- Meme coin and trend listings continue — WEEX listed a "2026" meme token tied to World Cup excitement on Solana, (WEEX)
- People compare Hermes and OpenClaw on skills, memory, and token burn. I compare them on one thing: which decisions each
- @omc345 @openclaw they didnt eat it's lunch. They beat it, took it's lunch koney and went to a strip club together inste
- @blknoiz06 check @Parad0x_Labs ca:8EeDdvCRmFAzVD4takkBrNNwkeUTUQh4MscRK5Fzpump VOOL → Local AI .null → Ownable identity — @blknoiz06 check @Parad0x_Labs ca:8EeDdvCRmFAzVD4takkBrNNwkeUTUQh4MscRK5Fzpump VOOL → Local AI .null → Ownable identity DNA x402 → Agent payments Dark Null → Privacy Liquefy → Data + verifiable audit...
- At this point just say Codex instead of OpenClaw.
- FastMetal の MCP サーバーを公開しました🔌 Claude Code / Codex / OpenClaw / Hermes に URL とキーを足すだけ。メインモデルは Claude のまま、詰まった箇所を別モデルに相談した — FastMetal の MCP サーバーを公開しました🔌 Claude Code / Codex / OpenClaw / Hermes に URL とキーを足すだけ。メインモデルは Claude のまま、詰まった箇所を別モデルに相談したり、8モデルに同時に投げて比べたり、画像を生成したり。 サーバー利用料は無料です👇