OpenClaw Newsletter - 2026-05-12
OpenClaw Newsletter
Tuesday, May 12, 2026Top Stories
- openclaw 2026.5.10-beta.5 — Changes - CI: add a non-blocking plugin-inspector-advisory artifact to Plugin Prerelease so release runs capture bundled plugin compatibility triage without changing the blocking gate. - Runtime/Fly:...
- openclaw 2026.5.10-beta.4 — Changes - CI: add a non-blocking plugin-inspector-advisory artifact to Plugin Prerelease so release runs capture bundled plugin compatibility triage without changing the blocking gate. - Runtime/Fly:...
- openclaw 2026.5.9-beta.1 — Changes - Chat commands: add /think default and /fast default to clear session overrides and inherit configured/provider defaults. (#79385) Thanks @VACInc. - Dependencies: refresh workspace...
- fix(slack): accept ISO 8601 timestamps in message tool after/before (#80835) — Summary The Slack message tool's after/before parameters silently returned empty arrays for ISO 8601 timestamp strings because Slack's conversations.history API expects epoch-seconds timestamps (.)....
- feat(subagents): forward toolsAllow from sessions_spawn — Summary - Add optional toolsAllow to sessions_spawn for native subagent runs. - Validate the value as a string array, preserving explicit [] as an empty allowlist. - Forward the allowlist through...
- ui(i18n): localize chat page slash commands and composer for zh-CN — Summary - Problem: Chat page slash command palette and composer had hardcoded English strings when using Simplified Chinese locale. - Why it matters: Non-English users see English labels,...
- String model config silently disables fallbacks (resolveAgentModelFallbackValues returns [] for strings) — Problem When agents.defaults.model is set as a string (e.g. "anthropic/claude-opus-4-7"), the function resolveAgentModelFallbackValues() in src/config/model-input.ts returns an empty array — silently...
- fix(telegram): add suppressReply support for registerCommand handlers — Summary When a plugin command handler delivers its own response via the Telegram Bot API and returns { suppressReply: true }, the channel adapter now skips the default "No response generated"...
- test: reproduce silent fallback no-reply drop — Summary - Adds a focused repro test for silent operational failure after model failover. - Covers the incident shape where a channel/group final reply is NO_REPLY but carries model-failover failure...
- [Feature]: Detect and recover from output truncation (stopReason:"length") in main agent sessions — Summary When the model hits its max_output_tokens limit mid-generation (especially during tool_use with large parameters), the agent runtime silently proceeds with truncated output. There is no...
- feat(embedded-runner): expose prep stage timings — Summary - expose existing embedded runner prep stage snapshots on attempt results and run meta - include prep stage timings on the agent_end hook event for plugin observability - add coverage for...
- CLI: bake git commit at build time so --version reports built commit not live HEAD [AI-assisted] — Fixes #68148. Summary Before: node dist/index.js --version reports the current value of .git/HEAD, which can drift past the commit the binary was actually built from. After: the short SHA is baked...
- fix(gateway): hold startup-gated requests at server until post-attach (closes #67160) — Summary - Problem: The openclaw-gateway startup gate from #65365 responds UNAVAILABLE to chat.history and models.list for the 8–15 s window between [gateway] ready and post-attach sidecar...
- mattermost: fix /oc_* slash command 503 from duplicate slash-state instance — Summary - Problem: Mattermost slash commands (/oc_model, /oc_status, /oc_help, …) always return 503 Service Unavailable with {"response_type":"ephemeral","text":"Slash commands are not yet...
- fix(daemon): add aliases/default/bin to Linux fnm PATH resolver (fixes #68169) — Fix issue #68169: Linux service PATH resolver now checks both aliases/default/bin (modern fnm) and current/bin (legacy fnm).
- doctor findOtherStateDirs scans /home/*/ — cross-user info leak and false positives on sibling backup dirs — findOtherStateDirs(stateDir) in dist/prompt-select-styled-.js scans /home// for .openclaw/ subdirs regardless of who's running. The relevant logic: js function findOtherStateDirs(stateDir) { const...
- fix(memory): apply decay to nested dated files — Summary - Fix temporal decay so it also applies to dated memory files stored in memory/ subdirectories. - Previously, the dated-memory path matcher only recognized top-level paths like...
- fix(ui): use i18n t() for Node option in exec approvals target select — What Fix hardcoded Node string in ui/src/ui/views/nodes-exec-approvals.ts that bypasses the i18n system. The translation key nodes.binding.node already exists in all locale files with proper...
- feat(daemon): use PowerShell Register-ScheduledTask for Windows auto-start — Problem 1. schtasks /Create has no built-in way to disable the default ExecutionTimeLimit, which can cause long-running tasks like the gateway to be terminated after 72 hours. 2. The previous code...
- [Bug]: Local assistant attachments shown as "Unavailable — Outside allowed folders" despite correct server config — Bug type Behavior bug (incorrect output/state without crash) Beta release blocker No Summary Local assistant media attachments (e.g. TTS audio in MEDIA: lines) render as \"Unavailable — Outside...
- Fix memory_search wiki fallback and plugin tool registry reuse — Summary - add a direct memory-wiki fallback for memory_search and memory_get when wiki corpus supplements are missing at runtime - avoid mutating global plugin memory state during plugin tool...
- fix(telegram): skip delete before non-empty command sync — Summary - skip deleteMyCommands() before syncing a non-empty Telegram command menu - preserve explicit deletes when syncing an empty menu - add targeted tests for both paths and stabilize the...
- fix: remove truncated preview from inbound system events — Summary Removes the truncated 160-char message preview from system event headers for Slack, MS Teams, and Mattermost channels. Problem System events were including a truncated preview of the message...
- [Bug]: Successful auto-compaction can still end in a 120s embedded timeout and generic `/new` fallback — Summary On a long-running embedded-agent session, OpenClaw can: 1. detect context overflow, 2. successfully auto-compact the session, 3. retry the prompt, 4. hit a 120s LLM idle timeout on the...
- [Bug]: bootstrap/reconcile and hot-cache policy can leave deferred compaction debt stranded — Bug type Behavior bug (incorrect runtime state without crash) Beta release blocker No Summary Context-engine / LCM maintenance can leave deferred compaction debt stranded in at least two ways: 1. A...
- gateway.auth.mode=none breaks internal RPC probe with 1008 device identity required — Summary When gateway.auth.mode is set to none, the Gateway's own internal RPC probe fails with: This causes a cascading failure chain: 1. All exec approval followup dispatches timeout after 60s 2....
- fix(memory): preserve surrogate pairs in chunker; sanitize embed inputs — Summary - Fix the outer coarse-split loop in chunkMarkdown so it does not bisect a UTF-16 surrogate pair (root cause of indexing stalls). - Add a defensive stripUnpairedSurrogates sanitizer at the...
- Plugin SDK: add text-only tool_result_before_model hook — Summary - Problem: plugins had no typed seam to canonicalize successful tool-result text before it became model-visible by default, so noisy tool output could leak into same-turn and future-turn...
- Feature request: Improve context pruning/trimming mechanism visibility and control — OpenClaw has internal context pruning/trimming to manage token limits, but users have limited visibility and control over this process. Current issues observed: 1. Context bloat from tool outputs:...
- feat(gateway): add audit logging for gateway tool calls — Summary Describe the problem and fix in 2–5 bullets: - Problem: gateway-originated tool calls did not have a centralized, append-only audit trail across direct /tools/invoke requests and...
- Control UI blank screen: unresolved bare module import for markdown-it-task-lists — Bug After upgrading to OpenClaw 2026.4.14, the Control UI serves a blank page on all browsers (including localhost on the Pi, incognito windows, etc.). Root Cause The built JS bundle...
- [Bug] openai-codex provider blocked by Cloudflare JS Challenge when accessed via proxy in mainland China — Environment - OpenClaw version: 2026.4.14 - OS: macOS 26.3 (arm64), Mac mini - Node.js: v22.22.2 - Network: Mainland China, traffic routed through VPN/proxy (Quantumult X) Problem The openai-codex...
- OpenClaw repo stats for 2026-05-12 — 371067 stars, 76736 forks, 2212 contributors
- OpenClaw sponsors (125 total) — 125 sponsors supporting OpenClaw
- test: wait for hydration before owner navigation
- fix: improve mobile detail tab targets
- ci: ignore overbroad tanstack history advisory
- fix: prepare clawhub 0.15.0
- Update styles.css
- Update .gitignore
- fix: show plugin settings action to managers (#2168)
- fix: show plugin settings action to managers (#2163)
- fix: clarify skill summary settings copy (#2162)
- docs: clarify pre-pr validation gates (#2161)
- fix: show full skill description on detail pages (#2160)
- fix: expand skill summary clamp (#2158)
- fix: polish dashboard artifact cards (#2157)
- fix: clamp skill detail summaries (#2156)
- fix: refresh skill star state (#2155)
- fix: prevent skill tabs from scrolling (#2154)
- fix: show downloads as primary adoption metric (#2153)
- test: drop removed rescan e2e probes (#2152)
- fix: refresh detail page loading skeletons (#2150)
- fix: show owner names when handles are hidden (#2151)
- Add skill: nevo-david/postiz
- Update install instructions
- openclaw-cli 2026.5.7 on Homebrew — Your own personal AI assistant
- sulpikar2/Qwen3.5-4B-Claude-4.6-Opus-Reasoning-openclaw
- 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 isn't fooling me. I remember MS-DOS
- Nanobot: Ultra-Lightweight Alternative to OpenClaw
- How to Add Tamper-Evident Audit Trails to Your OpenClaw Assistant — OpenClaw runs locally and acts on your behalf — so when something goes wrong, can you prove what happened? This walkthrough adds Ed25519-signed audit receipts to every tool call in 60 seconds.
- OpenClaw 4-Agent Fleet Public — With a Bug I Just Diagnosed — I'm publishing my full 24/7 AI agent fleet architecture. While doing so, I found a bug that quietly broke 2 of 3 agents for 20 days. Diagnosed and fixed from Gate C2 in 30 minutes.
- How we compare OpenClaw skills without collapsing everything into one fake score — If you're trying to decide which OpenClaw skills deserve attention first, a giant list doesn't help...
- We gave actual claws to Openclaw agent and it flies a drone now — A few weeks back, we posted a short demo of a drone following a car in peak SF traffic, controlled...
- Running a Personal AI Assistant for $0 - Part 1 - Architecture — Introduction A productivity tool that promised to change everything, charged monthly, and...
- 5 Things to Do After Installing OpenClaw (Before You Break It) — I've been running OpenClaw for a while now and I think it's pretty awesome. If you haven't heard of...
- Congrats to the OpenClaw Challenge Winners! — The results are in! We are thrilled to announce the winners of the OpenClaw Challenge. Over the past...
- Emails Are the New Markdown Files — Why EmailGPT May Be the Most Scalable Agent Architecture Hiding in Plain SightContinue reading on Medium »
- OpenClaw/Hermes/DeerFlow All Work Together — Connect to LinkMind in Minutes — Many developers juggle multiple AI agents at once, each running in its own silo: API keys scattered across projects, agents completely…Continue reading on Medium »
- Why Rate Limits Kill Developer Productivity (And How Proxy Pooling Fixes It) — If you’ve ever hit a 429 error mid-sprint with Claude API, you know the pain. This isn’t about cost — it’s about availability. Official…Continue reading on Medium »
- OpenClaw: Ngoding Cuma Lewat Prompt di Telegram? Emang Bisa? — Dunia coding sekarang lagi berubah cepat banget. Dulu orang belajar syntax satu per satu, debugging manual, buka Stack Overflow tiap 5…Continue reading on Medium »
- How We Orchestrate AI Agents in Production with Paperclip, OpenClaw, and Hermes — From the idea of a “central brain” to a Kubernetes-deployed multi-agent architecture a field report from BeezzonlineLab.Continue reading on Medium »
- Depth vs. Reach: Deconstructing the Architecture of Hermes Agent — The open-source AI agent ecosystem just hit a massive inflection point.Continue reading on Medium »
- I built an agentic CEO because I kept losing my own ideas — Three months in, I couldn’t have told you what I was building. I started with a clear, one-sentence idea. I could explain it at dinner. I…Continue reading on Altitudedp »
- We gave actual claws to openclaw agent and it flies a drone now — Openclaw agent controlling drone via MavlinkContinue reading on Medium »
- Dispositivo para verificar hornallas a gas con IA (Arduino UNO Q + OpenClaw) — En Argentina la mayoria de las hornallas funcionan a gas y es bastante común que las personas con trastornos obsesivos compulsivos…Continue reading on Medium »
- 8 Things Hermes Does That OpenClaw Can’t — Spend real time with both and you’ll notice something strange. They don’t feel like competitors. They feel like completely different tools…Continue reading on Medium »
- Apollo AI — Customizable interface for private, on-device AI model chatting, open source and closed source LLM connections, OpenRouter support, custom backend support, and unified access to language models...
- Maid (AI) — Maid is a cross-platform Flutter app for interfacing with GGUF / llama.cpp models locally, and with Ollama and OpenAI models remotely.
- $DOGE looking like a clean entry according to openclaw Was Made Using - @OpenClawAIX J71oozcoh6BcYjk14V7vjb1EcHk
- openclaw on $BIRDCLAW —momentum strong, risk low. easy Was Made Using - @OpenClawAIX CPCksBTG5N2Tp519i5mMnLt2Az2
- $PROG momentum score just came in hot on openclaw 🔥 Was Made Using - @OpenClawAIX 6AxUt7CycQcSwwPVex89rh4gw5SmaN
- openclaw on $BABYGOBLIN —momentum strong, risk low. easy Was Made Using - @OpenClawAIX FBAr3uXdSR9wfTHV6ve5zeUco
- openclaw on $BABYGOBLIN —momentum strong, risk low. easy ┃🧫┃ $BABYGOBLIN ┃💰┃ MC : $112.3K ┃🎯┃ TARGET MC : $1.4M ┃📈┃
- OpenClaw 版主：「因為水軍太多，我們禁止討論競爭對手。」 笑死。 背後有個很有趣的邏輯： 當一個產品需要透過「禁止批評」來維護形象， 說明它的用戶群太忠誠了，忠誠到願意當水軍。 而水軍多得讓版主必須禁止， 說明這個產品的好， — OpenClaw 版主：「因為水軍太多，我們禁止討論競爭對手。」 笑死。 背後有個很有趣的邏輯： 當一個產品需要透過「禁止批評」來維護形象， 說明它的用戶群太忠誠了，忠誠到願意當水軍。 而水軍多得讓版主必須禁止， 說明這個產品的好，已經不用多說了。 有個類比： 蘋果從來不禁止討論 Android。 https://t.co/zMnCQ1HQSW
- @icanvardar Codex is a code tool, openclaw is running tasks on your behalf, they're not really the same thing.
- openclaw called it on $BMNTP again. the model doesn't miss ┃🧫┃ $BMNTP ┃💰┃ MC : $32.7K ┃🎯┃ TARGET MC : $369.2K ┃📈┃ P
- $CROWDCAM looking like a clean entry according to openclaw Was Made Using - @OpenClawAIX FUkQZqqYSx4rgCRJfMoTyeF
- $BMNTP looking like a clean entry according to openclaw Was Made Using - @OpenClawAIX 6EQKNJD6KMTQv9KmhKDjs1jm1S
- openclaw flagged $Bear early. price is moving exactly as predicted Was Made Using - @OpenClawAIX 8BuGJvmzrtKg1Pq
- $Apple looking like a clean entry according to openclaw ┃🧫┃ $Apple ┃💰┃ MC : $253.9K ┃🎯┃ TARGET MC : $4.0M ┃📈┃ POTEN
- $unc is looking good on openclaw right now ngl Was Made Using - @OpenClawAIX ACtfUWtgvaXrQGNMiohTusi5jcx5RJf5zwu
- openclaw been right on $Peng twice already. trusting the scan Was Made Using - @OpenClawAIX CGMa4aA1F2Vfvuo5ndBR
- just checked $Tendies on openclaw—forecast looking clean 🦞 Was Made Using - @OpenClawAIX 3N2vjBgT2abbiGfiNQhRsoW
- $RKC signals are clean on openclaw. this one's on my radar 🦞 ┃🧫┃ $RKC ┃💰┃ MC : $3.5M ┃🎯┃ TARGET MC : $33.2M ┃📈┃ POT
- openclaw on $AGI —momentum strong, risk low. easy Was Made Using - @OpenClawAIX 4ZEzC3aX7yLv8VEBiuoT6PgEMPEoxGB7
- openclaw called $GKC perfectly last time. running the scan again ┃🧫┃ $GKC ┃💰┃ MC : $66.2K ┃🎯┃ TARGET MC : $835.2K ┃
- been watching $AGI through openclaw, momentum is building ┃🧫┃ $AGI ┃💰┃ MC : $349.0K ┃🎯┃ TARGET MC : $4.0M ┃📈┃ POTEN
- my openclaw scan on $SCAM just confirmed what i was thinking ┃🧫┃ $SCAM ┃💰┃ MC : $540.0K ┃🎯┃ TARGET MC : $3.3M ┃📈┃ P
Trending on X
- $DOGE looking like a clean entry according to openclaw Was Made Using - @OpenClawAIX J71oozcoh6BcYjk14V7vjb1EcHk
- openclaw on $BIRDCLAW —momentum strong, risk low. easy Was Made Using - @OpenClawAIX CPCksBTG5N2Tp519i5mMnLt2Az2
- $PROG momentum score just came in hot on openclaw 🔥 Was Made Using - @OpenClawAIX 6AxUt7CycQcSwwPVex89rh4gw5SmaN
- openclaw on $BABYGOBLIN —momentum strong, risk low. easy Was Made Using - @OpenClawAIX FBAr3uXdSR9wfTHV6ve5zeUco
- openclaw on $BABYGOBLIN —momentum strong, risk low. easy ┃🧫┃ $BABYGOBLIN ┃💰┃ MC : $112.3K ┃🎯┃ TARGET MC : $1.4M ┃📈┃
- OpenClaw 版主：「因為水軍太多，我們禁止討論競爭對手。」 笑死。 背後有個很有趣的邏輯： 當一個產品需要透過「禁止批評」來維護形象， 說明它的用戶群太忠誠了，忠誠到願意當水軍。 而水軍多得讓版主必須禁止， 說明這個產品的好， — OpenClaw 版主：「因為水軍太多，我們禁止討論競爭對手。」 笑死。 背後有個很有趣的邏輯： 當一個產品需要透過「禁止批評」來維護形象， 說明它的用戶群太忠誠了，忠誠到願意當水軍。 而水軍多得讓版主必須禁止， 說明這個產品的好，已經不用多說了。 有個類比： 蘋果從來不禁止討論 Android。 https://t.co/zMnCQ1HQSW
- @icanvardar Codex is a code tool, openclaw is running tasks on your behalf, they're not really the same thing.
- openclaw called it on $BMNTP again. the model doesn't miss ┃🧫┃ $BMNTP ┃💰┃ MC : $32.7K ┃🎯┃ TARGET MC : $369.2K ┃📈┃ P
- $CROWDCAM looking like a clean entry according to openclaw Was Made Using - @OpenClawAIX FUkQZqqYSx4rgCRJfMoTyeF
- $BMNTP looking like a clean entry according to openclaw Was Made Using - @OpenClawAIX 6EQKNJD6KMTQv9KmhKDjs1jm1S
- openclaw flagged $Bear early. price is moving exactly as predicted Was Made Using - @OpenClawAIX 8BuGJvmzrtKg1Pq
- $Apple looking like a clean entry according to openclaw ┃🧫┃ $Apple ┃💰┃ MC : $253.9K ┃🎯┃ TARGET MC : $4.0M ┃📈┃ POTEN
- $unc is looking good on openclaw right now ngl Was Made Using - @OpenClawAIX ACtfUWtgvaXrQGNMiohTusi5jcx5RJf5zwu
- openclaw been right on $Peng twice already. trusting the scan Was Made Using - @OpenClawAIX CGMa4aA1F2Vfvuo5ndBR
- just checked $Tendies on openclaw—forecast looking clean 🦞 Was Made Using - @OpenClawAIX 3N2vjBgT2abbiGfiNQhRsoW
- $RKC signals are clean on openclaw. this one's on my radar 🦞 ┃🧫┃ $RKC ┃💰┃ MC : $3.5M ┃🎯┃ TARGET MC : $33.2M ┃📈┃ POT
- openclaw on $AGI —momentum strong, risk low. easy Was Made Using - @OpenClawAIX 4ZEzC3aX7yLv8VEBiuoT6PgEMPEoxGB7
- openclaw called $GKC perfectly last time. running the scan again ┃🧫┃ $GKC ┃💰┃ MC : $66.2K ┃🎯┃ TARGET MC : $835.2K ┃
- been watching $AGI through openclaw, momentum is building ┃🧫┃ $AGI ┃💰┃ MC : $349.0K ┃🎯┃ TARGET MC : $4.0M ┃📈┃ POTEN
- my openclaw scan on $SCAM just confirmed what i was thinking ┃🧫┃ $SCAM ┃💰┃ MC : $540.0K ┃🎯┃ TARGET MC : $3.3M ┃📈┃ P
Releases
- openclaw 2026.5.10-beta.5 — Changes - CI: add a non-blocking plugin-inspector-advisory artifact to Plugin Prerelease so release runs capture bundled plugin compatibility triage without changing the blocking gate. - Runtime/Fly:...
- openclaw 2026.5.10-beta.4 — Changes - CI: add a non-blocking plugin-inspector-advisory artifact to Plugin Prerelease so release runs capture bundled plugin compatibility triage without changing the blocking gate. - Runtime/Fly:...
- openclaw 2026.5.9-beta.1 — Changes - Chat commands: add /think default and /fast default to clear session overrides and inherit configured/provider defaults. (#79385) Thanks @VACInc. - Dependencies: refresh workspace...
- OpenClaw repo stats for 2026-05-12 — 371067 stars, 76736 forks, 2212 contributors
- openclaw-cli 2026.5.7 on Homebrew — Your own personal AI assistant
Community
- fix(slack): accept ISO 8601 timestamps in message tool after/before (#80835) — Summary The Slack message tool's after/before parameters silently returned empty arrays for ISO 8601 timestamp strings because Slack's conversations.history API expects epoch-seconds timestamps (.)....
- feat(subagents): forward toolsAllow from sessions_spawn — Summary - Add optional toolsAllow to sessions_spawn for native subagent runs. - Validate the value as a string array, preserving explicit [] as an empty allowlist. - Forward the allowlist through...
- ui(i18n): localize chat page slash commands and composer for zh-CN — Summary - Problem: Chat page slash command palette and composer had hardcoded English strings when using Simplified Chinese locale. - Why it matters: Non-English users see English labels,...
- String model config silently disables fallbacks (resolveAgentModelFallbackValues returns [] for strings) — Problem When agents.defaults.model is set as a string (e.g. "anthropic/claude-opus-4-7"), the function resolveAgentModelFallbackValues() in src/config/model-input.ts returns an empty array — silently...
- fix(telegram): add suppressReply support for registerCommand handlers — Summary When a plugin command handler delivers its own response via the Telegram Bot API and returns { suppressReply: true }, the channel adapter now skips the default "No response generated"...
- test: reproduce silent fallback no-reply drop — Summary - Adds a focused repro test for silent operational failure after model failover. - Covers the incident shape where a channel/group final reply is NO_REPLY but carries model-failover failure...
- [Feature]: Detect and recover from output truncation (stopReason:"length") in main agent sessions — Summary When the model hits its max_output_tokens limit mid-generation (especially during tool_use with large parameters), the agent runtime silently proceeds with truncated output. There is no...
- feat(embedded-runner): expose prep stage timings — Summary - expose existing embedded runner prep stage snapshots on attempt results and run meta - include prep stage timings on the agent_end hook event for plugin observability - add coverage for...
- CLI: bake git commit at build time so --version reports built commit not live HEAD [AI-assisted] — Fixes #68148. Summary Before: node dist/index.js --version reports the current value of .git/HEAD, which can drift past the commit the binary was actually built from. After: the short SHA is baked...
- fix(gateway): hold startup-gated requests at server until post-attach (closes #67160) — Summary - Problem: The openclaw-gateway startup gate from #65365 responds UNAVAILABLE to chat.history and models.list for the 8–15 s window between [gateway] ready and post-attach sidecar...
- mattermost: fix /oc_* slash command 503 from duplicate slash-state instance — Summary - Problem: Mattermost slash commands (/oc_model, /oc_status, /oc_help, …) always return 503 Service Unavailable with {"response_type":"ephemeral","text":"Slash commands are not yet...
- fix(daemon): add aliases/default/bin to Linux fnm PATH resolver (fixes #68169) — Fix issue #68169: Linux service PATH resolver now checks both aliases/default/bin (modern fnm) and current/bin (legacy fnm).
- doctor findOtherStateDirs scans /home/*/ — cross-user info leak and false positives on sibling backup dirs — findOtherStateDirs(stateDir) in dist/prompt-select-styled-.js scans /home// for .openclaw/ subdirs regardless of who's running. The relevant logic: js function findOtherStateDirs(stateDir) { const...
- fix(memory): apply decay to nested dated files — Summary - Fix temporal decay so it also applies to dated memory files stored in memory/ subdirectories. - Previously, the dated-memory path matcher only recognized top-level paths like...
- fix(ui): use i18n t() for Node option in exec approvals target select — What Fix hardcoded Node string in ui/src/ui/views/nodes-exec-approvals.ts that bypasses the i18n system. The translation key nodes.binding.node already exists in all locale files with proper...
- feat(daemon): use PowerShell Register-ScheduledTask for Windows auto-start — Problem 1. schtasks /Create has no built-in way to disable the default ExecutionTimeLimit, which can cause long-running tasks like the gateway to be terminated after 72 hours. 2. The previous code...
- [Bug]: Local assistant attachments shown as "Unavailable — Outside allowed folders" despite correct server config — Bug type Behavior bug (incorrect output/state without crash) Beta release blocker No Summary Local assistant media attachments (e.g. TTS audio in MEDIA: lines) render as \"Unavailable — Outside...
- Fix memory_search wiki fallback and plugin tool registry reuse — Summary - add a direct memory-wiki fallback for memory_search and memory_get when wiki corpus supplements are missing at runtime - avoid mutating global plugin memory state during plugin tool...
- fix(telegram): skip delete before non-empty command sync — Summary - skip deleteMyCommands() before syncing a non-empty Telegram command menu - preserve explicit deletes when syncing an empty menu - add targeted tests for both paths and stabilize the...
- fix: remove truncated preview from inbound system events — Summary Removes the truncated 160-char message preview from system event headers for Slack, MS Teams, and Mattermost channels. Problem System events were including a truncated preview of the message...
- [Bug]: Successful auto-compaction can still end in a 120s embedded timeout and generic `/new` fallback — Summary On a long-running embedded-agent session, OpenClaw can: 1. detect context overflow, 2. successfully auto-compact the session, 3. retry the prompt, 4. hit a 120s LLM idle timeout on the...
- [Bug]: bootstrap/reconcile and hot-cache policy can leave deferred compaction debt stranded — Bug type Behavior bug (incorrect runtime state without crash) Beta release blocker No Summary Context-engine / LCM maintenance can leave deferred compaction debt stranded in at least two ways: 1. A...
- gateway.auth.mode=none breaks internal RPC probe with 1008 device identity required — Summary When gateway.auth.mode is set to none, the Gateway's own internal RPC probe fails with: This causes a cascading failure chain: 1. All exec approval followup dispatches timeout after 60s 2....
- fix(memory): preserve surrogate pairs in chunker; sanitize embed inputs — Summary - Fix the outer coarse-split loop in chunkMarkdown so it does not bisect a UTF-16 surrogate pair (root cause of indexing stalls). - Add a defensive stripUnpairedSurrogates sanitizer at the...
- Plugin SDK: add text-only tool_result_before_model hook — Summary - Problem: plugins had no typed seam to canonicalize successful tool-result text before it became model-visible by default, so noisy tool output could leak into same-turn and future-turn...
- Feature request: Improve context pruning/trimming mechanism visibility and control — OpenClaw has internal context pruning/trimming to manage token limits, but users have limited visibility and control over this process. Current issues observed: 1. Context bloat from tool outputs:...
- feat(gateway): add audit logging for gateway tool calls — Summary Describe the problem and fix in 2–5 bullets: - Problem: gateway-originated tool calls did not have a centralized, append-only audit trail across direct /tools/invoke requests and...
- Control UI blank screen: unresolved bare module import for markdown-it-task-lists — Bug After upgrading to OpenClaw 2026.4.14, the Control UI serves a blank page on all browsers (including localhost on the Pi, incognito windows, etc.). Root Cause The built JS bundle...
- [Bug] openai-codex provider blocked by Cloudflare JS Challenge when accessed via proxy in mainland China — Environment - OpenClaw version: 2026.4.14 - OS: macOS 26.3 (arm64), Mac mini - Node.js: v22.22.2 - Network: Mainland China, traffic routed through VPN/proxy (Quantumult X) Problem The openai-codex...
- OpenClaw sponsors (125 total) — 125 sponsors supporting OpenClaw
- test: wait for hydration before owner navigation
- fix: improve mobile detail tab targets
- ci: ignore overbroad tanstack history advisory
- fix: prepare clawhub 0.15.0
- Update styles.css
- Update .gitignore
- fix: show plugin settings action to managers (#2168)
- fix: show plugin settings action to managers (#2163)
- fix: clarify skill summary settings copy (#2162)
- docs: clarify pre-pr validation gates (#2161)
- fix: show full skill description on detail pages (#2160)
- fix: expand skill summary clamp (#2158)
- fix: polish dashboard artifact cards (#2157)
- fix: clamp skill detail summaries (#2156)
- fix: refresh skill star state (#2155)
- fix: prevent skill tabs from scrolling (#2154)
- fix: show downloads as primary adoption metric (#2153)
- test: drop removed rescan e2e probes (#2152)
- fix: refresh detail page loading skeletons (#2150)
- fix: show owner names when handles are hidden (#2151)
- Add skill: nevo-david/postiz
- Update install instructions
News
- sulpikar2/Qwen3.5-4B-Claude-4.6-Opus-Reasoning-openclaw
- 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 isn't fooling me. I remember MS-DOS
- Nanobot: Ultra-Lightweight Alternative to OpenClaw
- How to Add Tamper-Evident Audit Trails to Your OpenClaw Assistant — OpenClaw runs locally and acts on your behalf — so when something goes wrong, can you prove what happened? This walkthrough adds Ed25519-signed audit receipts to every tool call in 60 seconds.
- OpenClaw 4-Agent Fleet Public — With a Bug I Just Diagnosed — I'm publishing my full 24/7 AI agent fleet architecture. While doing so, I found a bug that quietly broke 2 of 3 agents for 20 days. Diagnosed and fixed from Gate C2 in 30 minutes.
- How we compare OpenClaw skills without collapsing everything into one fake score — If you're trying to decide which OpenClaw skills deserve attention first, a giant list doesn't help...
- We gave actual claws to Openclaw agent and it flies a drone now — A few weeks back, we posted a short demo of a drone following a car in peak SF traffic, controlled...
- Running a Personal AI Assistant for $0 - Part 1 - Architecture — Introduction A productivity tool that promised to change everything, charged monthly, and...
- 5 Things to Do After Installing OpenClaw (Before You Break It) — I've been running OpenClaw for a while now and I think it's pretty awesome. If you haven't heard of...
- Congrats to the OpenClaw Challenge Winners! — The results are in! We are thrilled to announce the winners of the OpenClaw Challenge. Over the past...
- Emails Are the New Markdown Files — Why EmailGPT May Be the Most Scalable Agent Architecture Hiding in Plain SightContinue reading on Medium »
- OpenClaw/Hermes/DeerFlow All Work Together — Connect to LinkMind in Minutes — Many developers juggle multiple AI agents at once, each running in its own silo: API keys scattered across projects, agents completely…Continue reading on Medium »
- Why Rate Limits Kill Developer Productivity (And How Proxy Pooling Fixes It) — If you’ve ever hit a 429 error mid-sprint with Claude API, you know the pain. This isn’t about cost — it’s about availability. Official…Continue reading on Medium »
- OpenClaw: Ngoding Cuma Lewat Prompt di Telegram? Emang Bisa? — Dunia coding sekarang lagi berubah cepat banget. Dulu orang belajar syntax satu per satu, debugging manual, buka Stack Overflow tiap 5…Continue reading on Medium »
- How We Orchestrate AI Agents in Production with Paperclip, OpenClaw, and Hermes — From the idea of a “central brain” to a Kubernetes-deployed multi-agent architecture a field report from BeezzonlineLab.Continue reading on Medium »
- Depth vs. Reach: Deconstructing the Architecture of Hermes Agent — The open-source AI agent ecosystem just hit a massive inflection point.Continue reading on Medium »
- I built an agentic CEO because I kept losing my own ideas — Three months in, I couldn’t have told you what I was building. I started with a clear, one-sentence idea. I could explain it at dinner. I…Continue reading on Altitudedp »
- We gave actual claws to openclaw agent and it flies a drone now — Openclaw agent controlling drone via MavlinkContinue reading on Medium »
- Dispositivo para verificar hornallas a gas con IA (Arduino UNO Q + OpenClaw) — En Argentina la mayoria de las hornallas funcionan a gas y es bastante común que las personas con trastornos obsesivos compulsivos…Continue reading on Medium »
- 8 Things Hermes Does That OpenClaw Can’t — Spend real time with both and you’ll notice something strange. They don’t feel like competitors. They feel like completely different tools…Continue reading on Medium »
- Apollo AI — Customizable interface for private, on-device AI model chatting, open source and closed source LLM connections, OpenRouter support, custom backend support, and unified access to language models...
- Maid (AI) — Maid is a cross-platform Flutter app for interfacing with GGUF / llama.cpp models locally, and with Ollama and OpenAI models remotely.
Don't miss what's next. Subscribe to Openclaw Newsletter: