- fix(memory): rank filename and path matches — What Problem This Solves memory_search indexed only chunk bodies. Exact filename and date-stem queries such as 2026-06-17-1649 could therefore behave like vector-only semantic searches and omit the...
- fix(agents): preserve Anthropic Retry-After for auto-retry backoff — What Problem This Solves Fixes #103849. Anthropic Messages transport dropped HTTP status and Retry-After on error responses, so auto-retry backoff ignored the server cooldown and could retry too...
- chore(discord): restore proxy multipart test on Node 24 — What Problem This Solves Resolves a clean-main Node 24 Testbox failure in the Discord proxy multipart regression test. The test mixed Node's global FormData with the separately installed...
- [Feature]: clean up empty openclaw-staged-* inbound media staging directories automatically — Summary When inbound media is staged into the agent workspace, OpenClaw creates per-run directories under media/inbound/openclaw-staged-. Over time these can accumulate as empty leftover directories....
- fix(brave): forward the execution abort signal to web search HTTP requests — What Problem This Solves Cancelling an agent run while a Brave web_search request is pending leaves that HTTP request running. The parent execution signal reaches the generic web-search runtime, but...
- fix(line): use the real LINE message id in push receipts — What Problem This Solves Fixes an issue where LINE push sends report a placeholder message id instead of the real one. pushLineMessages returned a hardcoded messageId: "push" in the send receipt, so...
- chore(tooling): typecheck scripts/** with a dedicated tsgo lane — Closes #104313 What Problem This Solves Resolves a coverage gap where scripts/** TypeScript files — including release checks, release verifiers, and build-pipeline scripts — were never typechecked....
- fix(maint): reuse recent same-PR hosted gates — What Problem This Solves The landing wrapper only reused recent hosted-gate evidence after its own controlled rebase. Ordinary follow-up pushes therefore waited for an entirely new gate cohort even...
- fix(gateway): add per-request cancellation to prevent late node.invoke dispatch — What Problem This Solves A pending plugin approval can execute dangerous node.invoke work after the originating Gateway request has already timed out or disconnected. This is a P1 security...
- fix(web-search): align credential presence with usable providers — What Problem This Solves Fixes web-search credential presence detection so the audit only reports usable web-search credentials after applying plugin policy, selected provider scoping, and...
- fix: stop Gateway without source rebuild — Summary - stop the managed Gateway through the existing built dist/index.js - prevent source launchers from auto-building stale output before the stop is dispatched Proof - live updater output showed...
- fix(firecrawl,brave,web-search,web-fetch,models-config): use own-property checks for config field lookups — What Problem This Solves All runtime config lookup paths use "" in to check whether a config field is present. The in operator traverses the prototype chain, so an inherited property (e.g. via...
- fix: protect SMS webhook callback quota [AI] — What Problem This Solves Fixes an issue where invalid SMS webhook posts could consume the same inbound callback quota used by signed Twilio callbacks when deployments receive traffic through a shared...
- fix(discord): add timeouts to directory lookup requests — What Problem This Solves Fixes an issue where Discord directory, user allowlist, and channel allowlist lookup requests could hang indefinitely when Discord API fetches stalled before headers or...
- fix(mcp): map disabled flag to enabled in config normalization — What Problem This Solves Resolves #103954 When an MCP server entry in openclaw.json has disabled: true, the gateway silently ignores it and spawns the processes anyway. This PR fixes the bug by...
- fix(ci): sync Control UI raw-copy baseline — What Problem This Solves Current main fails the Control UI i18n raw-copy check after the BTW composer added intentional labels/text without updating the tracked baseline. This blocks unrelated pull...
- feat(skills): preview proposals before approval — Closes #103871 Summary - add a read-only review action to skill_workshop - render create proposals as their complete canonical applied skill and update proposals as unified diffs - bound previews...
- feat: add interactive launcher menu as thin wrapper over existing CLI — What Problem This Solves Users need a simple interactive menu to launch OpenClaw without remembering CLI commands. This PR adds a thin wrapper launcher that provides a friendly menu UX while...
- ci(release): preserve Telegram launcher stage — What Problem This Solves The exact-merge Telegram release canary now fails quickly, but the launcher reports every nested mount-namespace failure as enter-mount-namespace. That hides whether the...
- fix(telegram): add timeouts to getChat lookup requests — What Problem This Solves Fixes an issue where Telegram @username target resolution could wait indefinitely when the Bot API getChat request stalled before headers or while reading the response body....
- fix(discord): replace gateway READY retry fixed backoff with exponential backoff — What Problem This Solves Discord gateway READY wait retries use a fixed 2-second backoff (DISCORD_GATEWAY_READY_RETRY_BACKOFF_MS = 2_000). When the gateway is temporarily unstable (e.g. network...
- feat(gateway): add operator killswitch to pause/abort agent runs — What Problem This Solves There is currently no way to immediately pause a misbehaving agent without a full openclaw daemon service stop, which also kills every channel transport — so an operator away...
- fix(discord): guard progress draft collapse to prevent re-collapse on cleaned stream — What Problem This Solves After v2026.6.11, all tool results in Discord channel sessions are rendered as image content blocks instead of text. The model receives "(see attached image)" placeholders...
- feat: show build identity chip (commit + age + branch) in sidebar footer — Closes #104338 What Problem This Solves When running OpenClaw from source (the common maintainer/operator setup), there is no glanceable way to tell which build the Control UI is running: which...
- fix(slack): suppress progress chrome sends — Summary - detect progress-only Slack tool chrome payloads with leading tool emoji plus command/status body - convert thread-scoped chrome to reactions instead of chat.postMessage - fail closed by...
- improve(ci): overlap Kova and source performance probes — Fixes #104346 What Problem This Solves Resolves a performance-workflow problem where independent Kova validation and OpenClaw source probes ran serially on one runner, making the 8m44s workflow pay...
- chore(autoreview): sync hardened org review defaults
- feat(workspaces): add agent-composable Workspaces — Related: #101093 Supersedes the stacked implementation in #101094, #101097, and #101098 while preserving @100yenadmin's original product direction and contributor credit. What Problem This Solves...
- refactor(agents): converge CLI tool terminal reason onto run-termination — What Problem This Solves The one-shot and live CLI runners each derived the same terminal decision — timed_out / cancelled / failed for tools when a run ends — with different logic: -...
- fix(discord): add timeouts to PluralKit lookup requests — What Problem This Solves Discord PluralKit lookups are optional inbound message enrichment, but the PluralKit fetch and response body read did not have a deadline. A slow third-party API response...
- OpenClaw repo stats for 2026-07-11 — 382555 stars, 80298 forks, 2857 contributors
- OpenClaw sponsors (106 total) — 106 sponsors supporting OpenClaw
- fix(search): gate exact-match rank on trust and order tiers by adoption (#3058)
- fix: scale sanitized snapshot processing
- feat: adopt shared design system across ClawHub UI (#3026) — chore: install OpenClaw design system feat: adopt shared design system palette chore: automate design system updates feat: add weekly design system audit fix: prevent mobile skills tab overlap fix:...
- Add ClawHub overview link to README — Added a link to Trent AI's ClawHub overview for more context.
- Revise trentclaw section in README — Updated trentclaw description and image in README.
- openclaw/openclaw on Docker Hub — Official openclaw/openclaw docker images.
- sulpikar2/Qwen3.5-4B-Claude-4.6-Opus-Reasoning-openclaw
- sunkencity/qwen25-3b-openclaw — 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...
- Tell HN: Anthropic no longer allowing Claude Code subscriptions to use OpenClaw
- OpenClaw is what Apple intelligence should have been
- StepFun 3.5 Flash is #1 cost-effective model for OpenClaw tasks (300 battles)
- How to Run AI Locally: Models, Hardware, and Real-World Speed
- The Best Hardware for OpenClaw in 2026 (Jetson, Mini PC, or DIY?)
- Onboard OpenClaw agents with IdentyClaw Passport: A2A, webhooks, and multi-tenant collaboration — If you run more than one OpenClaw agent — or agents that talk to peers outside your VPC — you have...
- OpenClaw Hardware Requirements: What You Actually Need to Run It 24/7
- How to Set Up Multiple Macs for Always-On AI Agents — The multi-Mac setup I wish someone had handed me when I was still screen-sharing into a headless computer. I basically never SSHed into my…Continue reading on Coding Nexus »
- Helpful Skills or Hidden Payloads? Bitdefender Labs Dives Deep into the OpenClaw Malicious Skill Trap — With hundreds of malicious OpenClaw skills blending in among legitimate ones, manually reviewing every script or command isn’t realistic — especially when skills are designed to look helpful and...
- RT @elonmusk: Grok 4.5 on OpenClaw
- RT @MoltenRockAI: Claude Desktop, Cursor, OpenClaw. If you use AI agents, MoltenMail gives them access to your email inb
- RT @elonmusk: Grok 4.5 on OpenClaw
- RT @anton_bt: THIS GUY TURNED A MAC MINI INTO THE BRAIN FOR AN OPENCLAW AGENT TEAM RUNNING CLIENT AND BUSINESS WORK The
- @TheTotallyRed openclaw guy
- @thegrowthyard I got nikita, openclaw guy and f dot inc guy
- RT @LilyDash: @KongBTC @amandaterry @openclaw @FutureCaribb @the_IDB @miaamormottley @KiteVC @ACTAIVentures @OnChainMonk
- RT @MoltenRockAI: Claude Desktop, Cursor, OpenClaw. If you use AI agents, MoltenMail gives them access to your email inb
- @GithubProjects https://t.co/xgHRNXUihA i got 3 research write ups and many fun projects like my own openclaw for chrom — @GithubProjects https://t.co/xgHRNXUihA i got 3 research write ups and many fun projects like my own openclaw for chrome and a rust based inference engine for language models
- RT @QT9277: AI Agent 上网调研神器 Agent-Reach，开源免费！ 适配 OpenClaw、Hermes Agent、CodeX 全系列智能体 解决 X 收费 API、B 站小红书风控、网页抓取付费等各类痛点 一
- @ALTIC_DEV https://t.co/qAwoasAyNZ it’s what things like openclaw and Hermes use at their core before adding bloat
- @michael_chomsky Also, the new T3 Code mobile app looks like it slaps! And Conductor shipping a mobile app this week! An
- RT @elonmusk: Grok 4.5 on OpenClaw
- RT @Dinosn: Researcher Details WhatsApp-to-Host Attack Chain Using Three OpenClaw Flaws https://t.co/dBxZYYhkxl
- RT @gregisenberg: Grok 4.5 might be the BEST model to run inside Hermes or OpenClaw RIGHT NOW. I've been sleeping on Gr
- @chunxiangai 是让 OpenClaw 调试了半年，终于把这条推文发出来了吗
- RT @gregisenberg: Grok 4.5 might be the BEST model to run inside Hermes or OpenClaw RIGHT NOW. I've been sleeping on Gr
- 7/7 En marge : mue MCP le 28/07 (stateless, OAuth 2.1), Tencent Hy3 candidat OpenClaw (Apache 2.0), OmniVoice 0.2.0 TTS — 7/7 En marge : mue MCP le 28/07 (stateless, OAuth 2.1), Tencent Hy3 candidat OpenClaw (Apache 2.0), OmniVoice 0.2.0 TTS 40x RTF, Claude Code v2.1.206. Une maison antifragile lit les brouillons de...
- 6/7 Nuit du samedi levant, 00:32. OpenClaw + Dante basculés sur ChatGPT gpt-5.5 via broker CLIProxyAPI. DeepSeek gardé e — 6/7 Nuit du samedi levant, 00:32. OpenClaw + Dante basculés sur ChatGPT gpt-5.5 via broker CLIProxyAPI. DeepSeek gardé en secours. Une chaîne de fallback vérifiée avant la prochaine panne, c'est ça,...