- openclaw 2026.5.22 — 2026.5.22 Changes - Gateway/perf: reuse process-stable channel catalog reads, avoid repeated bundled-channel boundary checks, and rotate gateway watch CPU profiles so benchmark runs do not accumulate...
- openclaw 2026.5.22-beta.1 — 2026.5.22 Changes - Docs: clarify README onboarding and Gateway startup paths, WhatsApp QR/408 recovery, cron output language prompts, skill advanced features, gateway upstream 403 troubleshooting,...
- fix(agents): strip plaintext model provider keys — Summary - strip plaintext provider apiKey values after merge-mode preservation, source-managed enforcement, and provider compatibility transforms so old models.json keys cannot be reintroduced into...
- Recover DeepSeek DSML tool calls — Fixes #85918. Summary - recover complete DeepSeek DSML tool_calls / tool_call / function_calls text blocks on the OpenAI completions stream into synthetic toolCall blocks - keep the existing native...
- Feature Request: Automated Session Memory Preservation & Synthesis — Summary Add automatic session memory preservation when /new (session reset) is executed, enabling continuous learning across sessions. Problem Currently, when a session resets (/new), all context is...
- (fix) gateway restart health timeout — Summary Describe the problem and fix in 2–5 bullets: - Problem: daemon status/restart reachability checks were using the heavier status RPC through the generic gateway call path, which is a poor...
- feat(exec): per-host allowlists via tools.exec.allowedHosts — Summary - Problem: tools.exec.host is a hard boundary — an agent configured with host: "gateway" cannot run commands in the sandbox, and vice versa. - Why it matters: agents that need restricted...
- fix(ui): show failed tool results as errors — Summary - Problem: failed tool results in chat could look like successful tool output with a checkmark. - Solution: propagate explicit isError / is_error flags into tool cards and add bounded...
- fix(telegram): suppress message-tool-only progress leaks — Summary Fixes a Telegram delivery leak where message-tool-only turns could still emit text-only tool/status progress as normal chat messages when verbose progress was enabled. In Telegram chats,...
- fix(install): validate downloaded scripts before execution — Problem Four code paths in scripts/install.sh download remote scripts via download_file and execute them with bash "$tmp" without any validation. While HTTPS/TLS1.2 protects the transport, truncated...
- Prefer Talk source-reply final text — Fixes #85275. Summary - keep the Talk consult listener active when the first matching chat final has no speakable text - prefer the later canonical final text that the Control UI displays for...
- Control UI: Model provider display shows original config value instead of normalized value (e.g., 'bailian' vs 'zai') — Describe the bug The Control UI displays the original configuration value for model providers instead of the normalized value used internally, causing confusion when switching between providers. To...
- [Feature]: Non-interactive command to use Codex OAuth for model provider — Summary Add non-interactive way to setup model provider using OpenAI Codex OAuth (e.g. just by supplying something like OPENAI_CODEX_OAUTH_CALLBACK_URL in the env) Problem to solve I have seen the...
- Telegram: add reaction notification wake policy — Summary - Problem: Telegram reaction events were queued as passive context, so reaction-only inputs did not wake the routed agent session. - Why it matters: operators may want selected notification...
- fix(slack): keep DM thread turns out of active steering — Summary - Preserve OpenClaw's stable Slack DM SessionKey contract for top-level DMs and ordinary DM thread replies. - Track routed Slack thread ids on active reply operations, including...
- [Bug]: 2026.5.22 Discord takes a long time to respond — Bug type Regression (worked before, now fails) Beta release blocker No Summary After upgrading to 2026.5.22 from 2026.5.18, Discord now takes a 30 seconds before the agent even "agent is typing" in...
- Add auth sync command or auto-sync for API key rotation across agents — Summary When updating env.ANTHROPIC_API_KEY via openclaw config set, the change doesn't propagate to agent auth-profiles.json files, causing silent 401 errors. Problem to solve Currently, when a user...
- Dockerfile: simplify multi-stage build — - Reduce from 5 stages to 2 (build + runtime) - Remove retry loops for Bun and Corepack installs - Hardcode bookworm-slim runtime (removes variant switching) - Inline extension copy (removes ext-deps...
- refactor: internalize OpenClaw agent runtime — Summary - rename the former Pi agent runtime/package surfaces to OpenClaw-owned agent runtime names - internalize the runtime packages into core and wire the embedded OpenClaw runtime through the...
- feat(android): add Wear OS app MVP — Summary > This is the follow up PR of #46279 > It was heavily written with the help of Gemini 3.1 Pro (mainly for building UI) and Codex GPT 5.4, including lots of review iterations. > During the...
- [Feature Request] Add streamTo="parent" support for runtime="subagent"
- Reminder guard fires false positives on memory-related "I'll remember" statements — Description The agent-runner-reminder-guard appends a visible disclaimer to agent responses when the model says things like "I'll remember" without scheduling a cron job. However, the regex pattern...
- feat(ios): Siri App Intent integration — native "Hey Siri, ask OpenClaw..." support — Summary The iOS app currently lacks native Siri App Intent support. Users can work around this with Shortcuts + Webhook, but there's no deep integration where Siri can handle parameters, stream...
- Agents: tighten tool name trust and preflight tool collisions — Summary - Problem: tool-result local media trust relied on normalized names, so external tools could collide with built-in names and inherit local path passthrough. - Why it matters: that made it...
- fix(doctor): show per-step progress spinners during update — Summary - Problem: openclaw doctor offers to update before running checks, but the update (fetch/rebase/build/ui:build/doctor) can take several minutes with zero progress indication — just a static...
- feat(exec): verified-human instruction provenance for sensitive targets — Summary Add a separate, gateway-enforced policy layer for instruction provenance on sensitive side-effect actions. CaMeL / taint provenance addresses the question: - Did untrusted content influence...
- Node host: no SSH transport support + challenge auth fails behind reverse proxy — Summary openclaw node run cannot connect to a remote gateway behind a TLS reverse proxy (nginx). The gateway's built-in SSH transport (gateway.remote.transport: "ssh") works for the gateway client...
- Reply generated but not delivered when compaction triggers session rollover — Bug Description When a session approaches context limits (~85% usage) and auto-compaction triggers with willRetry=true (overflow recovery), the LLM generates a reply that is visible in the session...
- Feature Request: recursive subagent listing (--depth / --recursive) — Problem With maxSpawnDepth: 2, we use the orchestrator pattern: main agent -> project manager (depth-1) -> worker experts (depth-2). The main agent needs visibility into depth-2 workers to monitor...
- feat: add user-specific memory isolation — Summary Adds user-specific memory isolation to OpenClaw, enabling per-user memory files in multi-user environments. Changes Core Feature - MemoryIsolationConfig type added to...
- Clamp fetch timeout timers — Summary - Clamp buildTimeoutAbortSignal delays through the existing resolveSafeTimeoutDelayMs helper before arming Node timers. - Add regression tests at both the shared signal builder and public...
- OpenClaw repo stats for 2026-05-24 — 374262 stars, 77834 forks, 2349 contributors
- OpenClaw sponsors (118 total) — 118 sponsors supporting OpenClaw
- fix: increase security scan worker throughput (#2389)
- fix: let codex adjudicate scan evidence (#2388)
- feat: support package security rescans (#2387)
- fix: streamline security audit agentic lanes (#2386)
- fix: require scan worker artifact reads (#2385)
- fix: make security scan worker dispatchable (#2384)
- feat: use SkillSpector for agentic risk findings (#2383)
- fix: point docs auth to canonical host
- Enhance README with self-hosting tips — Added tips for self-hosting OpenClaw Docker image and skills volume management.
- update deployment section
- update hero
- update contributing guide
- openclaw v2026.5.22 on npm — Multi-channel AI gateway with extensible messaging integrations
- openclaw-cli 2026.5.19 on Homebrew — Your own personal AI assistant
- mradermacher/Medina-Qwen3.5-27B-OpenClaw-Uncensored-i1-GGUF
- 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...
- Nanobot: Ultra-Lightweight Alternative to OpenClaw
- OpenClaw’s memory is unreliable, and you don’t know when it will break
- The Multi-Agent Framework I Actually Use (And Why I Stopped Using the Others) — I went through the same evaluation you're going through right now. LangGraph vs CrewAI vs OpenAI...
- Wiring MCP Into My Fitness Tracker — and Asking OpenClaw About My Last Workout — I built a Model Context Protocol server into the fitness tracker I vibe coded a year ago, wired it through Vercel and Coder workspaces, and ended the afternoon asking my Discord bot what my last...
- Delegation vs Collaboration vs Asking — The Four AI Work Modes Nobody Talks About — Microsoft's Worklab just published new research that will quietly reshape how you think about using...
- Openclaw to meta connection through Composio — Composio is the state of the art tool to do that, I have tried multiple and none was that good and responsive and free ;DContinue reading on Medium »
- Building Real AI Agents with OpenClaw (Free HandBook) — The Complete Engineering Guide to Autonomous AI Systems, Multi-Agent Architectures, and Production DeploymentContinue reading on Medium »
- How to Run OpenClaw with Open-Source Models — Run OpenClaw assistant through alternative LLMsContinue reading on Towards AI »
- How to connect Openclaw to Reddit — OpenClaw is an AI agent you can run locally and connect to tools, apps, and workflows.Continue reading on Medium »
- How to OpenClaw with Hostinger — Powerful 1 Click Setup Guide for Fast AI Agent Deployment — How to OpenClaw with Hostinger using a powerful 1 click setup that keeps your AI agent online 24/7 with built in web scraping, Telegram…Continue reading on Medium »
- Telefonumdan Bir Mesaj Attım, 10 Dakika Sonra Yayında Olan Mobil Uygulamama Yeni Bir Özellik… — Kendi sunucumda 7/24 çalışan bir AI agent, bir Telegram konuşması ve 4 dolarlık VPS. Bilgisayar yok. IDE yok. Sadece orkestrasyon.Continue reading on Medium »
- I Found the AI Tool That Actually Does the Work — Not Just Answers Questions — OpenClaw Platinum Bundle is the closest thing to hiring a full-time AI employee — here’s why marketers and entrepreneurs are going all in.Continue reading on Medium »
- Security, transparency, and execution efficiency are becoming the core foundations of financial infrastructure in the AI
- RT @aronhouyu: Ai领域最值得关注的英文账号： @karpathy = 大语言模型之王 @steipete = OpenClaw 创始人 @gregisenberg = 创业点子王 @rileybrown = VibeCod — RT @aronhouyu: Ai领域最值得关注的英文账号： @karpathy = 大语言模型之王 @steipete = OpenClaw 创始人 @gregisenberg = 创业点子王 @rileybrown = VibeCode 王 @jackfriks = 一人…
- openclaw on Poor —momentum strong, risk low. easy ┃🧫┃ $Poor ┃💰┃ MC : $518.4K ┃🎯┃ TARGET MC : $7.5M ┃📈┃ POTENTIAL :
- openclaw flagged $Poor as a strong hold. not surprised $Poor MC $518.4K → $7.5M (+1351%) 24h vol $2.4M Was Made Us
- openclaw called it on $Poor again. the model doesn't miss Was Made Using - @OpenClawAIX T1YZcgtG7Co3ZzFyZn5K9CjD
- my openclaw analyst flagged $ZECBank as a buy. noted 🦞 Was Made Using - @OpenClawAIX GyLh2TQD8CeJeCwpeZikD13kJob
- $EARL momentum score just came in hot on openclaw 🔥 $EARL MC $7.5K → $86.3K (+1044%) 24h vol $123.3K Was Made Usin
- RT @agentlayer_ai: v0.1.24 is live now Solana wallet onboarding is now much simpler One command installs the openclaw — RT @agentlayer_ai: v0.1.24 is live now Solana wallet onboarding is now much simpler One command installs the openclaw wallet runtime, wir…
- openclaw on $POOR —momentum strong, risk low. easy $POOR current → $4.6K target → $71.6K vol 24h → $123.0K
- openclaw flagged POOR as a strong hold. not surprised ┃🧫┃ $POOR ┃💰┃ MC : $4.6K ┃🎯┃ TARGET MC : $71.6K ┃📈┃ POTENTIAL
- RT @codewithimanshu: Make money with OpenClaw in 7 days. My OpenClaw Made $113,73 Autonomously and Now I’m Scared. You
- OpenClawの導入代行サービスが一般向けに始まったらしい。法人限定だったものが個人でも利用できるようになったとのこと。初期設定やセキュリティ、コスト最適化まで含め、1時間で環境構築するという。
- RT @yasinaktimur: Takip etmeniz gereken 16 AI figürü. @karpathy = LLMs @steipete = openclaw kurucusu @gregisenberg
- openclaw flagged $Artcoin early. price is moving exactly as predicted Was Made Using - @OpenClawAIX BqyTcTLht1Gr
- openclaw called $Artcoin perfectly last time. running the scan again $Artcoin MC $2.4K → $35.7K (+1414%) 24h vol $10
- $Cancercoin looking like a clean entry according to openclaw Was Made Using - @OpenClawAIX H2aTSa94AcWVvPFdos6YH
- RT @aronhouyu: Ai领域最值得关注的英文账号： @karpathy = 大语言模型之王 @steipete = OpenClaw 创始人 @gregisenberg = 创业点子王 @rileybrown = VibeCod — RT @aronhouyu: Ai领域最值得关注的英文账号： @karpathy = 大语言模型之王 @steipete = OpenClaw 创始人 @gregisenberg = 创业点子王 @rileybrown = VibeCode 王 @jackfriks = 一人…
- openclaw called nigga perfectly last time. running the scan again ┃🧫┃ $nigga ┃💰┃ MC : $2.4K ┃🎯┃ TARGET MC : $35.7K
- RT @uniswap12: 不用再花299请人卸载了，这里送你OpenClaw 彻底卸载教程（Windows / macOS） OpenClaw最近的爆火，让身边的很多人都患上了FOMO（错失恐惧症）身边不少人跟风“养龙虾”，结果没两天
- openclaw called $ATM bullish. i'm listening Was Made Using - @OpenClawAIX w9LTBBFqjcADdw52oLKjUUd7qimbBaeY68CFae