OpenClaw Newsletter - 2026-04-20
OpenClaw Newsletter
Monday, April 20, 2026Top Stories
- feat(plan-mode): full rollout + iter-1/2/3 hardening (consolidates PRs A/B/C/D/E/F/7/8/10/11) — Plan mode — full rollout + iter-1/2/3 hardening (consolidated stack of 10 PR's & 300+ Code Reviews) Executive summary This PR ships plan mode across OpenClaw: an opt-in, per-session workflow where...
- fix(cron): retry transient recurring failures before next schedule — Summary This changes recurring cron jobs to retry transient failures on the configured retry backoff window before falling back to their next natural schedule. Previously, transient failures for...
- fix(logging): add sink-level log redaction gate — Summary - Problem: log redaction was applied inconsistently at logging call sites, so file logs and external transports could still persist raw secrets even when some console-facing paths were...
- fix(ui): populate context-window cache from gateway model catalog — Summary - Problem: The context window usage meter in the Control UI always shows 0% for Anthropic/Claude models, even though session_status correctly reports real usage (e.g. "Context: 81k/200k...
- browser: support per-profile headless override — Summary - add browser.profiles..headless to the browser profile config schema - make resolveProfile() compute a per-profile headless value that falls back to global browser.headless - use the...
- fix(tasks): add tolerance for startedAt before createdAt in task audit — Summary openclaw tasks audit reports false-positive inconsistent_timestamps warnings on the majority of tasks (both succeeded and failed), making the audit output noisy and hiding real issues. Root...
- Qwen: qwen3.6-plus is blocked on Coding Plan Pro endpoint despite being officially supported — Problem OpenClaw hardcodes a filter that blocks qwen3.6-plus when the configured Qwen provider uses the Coding Plan endpoint (https://coding.dashscope.aliyuncs.com/v1), even though Alibaba Cloud...
- fix(acp): return SESSION_BUSY error when concurrent calls hit active session — When multiple concurrent openclaw agent --session-id calls target the same session while a turn is in progress, callers could be silently dropped. This fix checks for an active turn before entering...
- Yuanbo — Summary Describe the problem and fix in 2–5 bullets: If this PR fixes a plugin beta-release blocker, title it fix(): beta blocker - and link the matching Beta blocker: - issue labeled beta-blocker....
- fix(gateway): SIGKILL stale process when graceful shutdown times out on restart/stop — Problem Fixes #43960. restartLaunchAgent() already called waitForPidExit() before bootstrapping the new gateway, but if the old process was still alive after the 10-second SIGTERM window, the...
- fix(exec): surface signal-killed subprocesses as structured errors instead of silent undefined — Problem When a subprocess is killed by an OS signal (SIGKILL from OOM pressure, macOS debug.swd_panic=1 watchpoints, etc.) Node.js delivers exit(null, "SIGKILL") to the exit handler. The previous...
- fix: add logging for 4xx hook gateway errors — Summary Adds logging for all 4xx error responses in the hook gateway handler to improve debuggability. Root Cause When hook POSTs fail validation (HTTP 400), no trace is left in the gateway journal....
- Feature/add wecom channel — Summary - Problem: OpenClaw lacks WeCom channel support; users cannot connect to WeCom via the official WeCom Bot API. - Why it matters: WeCom is the dominant enterprise communication platform in...
- fix(hooks): preserve messageProvider in compaction and reset contexts — Summary Preserve messageProvider in before_compaction and before_reset hook contexts so plugins do not have to fall back to unknown on valid channel-scoped sessions. This fixes three concrete paths:...
- feat(agents): forward prompt_cache_key on openai-completions transport when compat.supportsPromptCacheKey is set — Summary buildOpenAIResponsesParams forwards prompt_cache_key: options?.sessionId on every Responses API turn (src/agents/openai-transport-stream.ts:794), and parseTransportChunkUsage already reads...
- fix: recover reasoning-only Ollama turns — Summary - allow Ollama runs to use the same incomplete-turn recovery guards as GPT-5 strict-agentic runs - retry reasoning-only and empty-response Ollama turns instead of surfacing them as unusable...
- Add per-session delete/archive action button in Control UI Sessions view — Currently, deleting sessions in the Control UI requires selecting checkboxes and using the bulk delete bar. This works for batch operations but is cumbersome for quick single-session actions....
- feat(qqbot): extract self-contained engine/ architecture with QR-code onboarding, approval handling — Summary - Problem: The QQBot built-in plugin (extensions/qqbot) had all gateway logic in a single 1500+ line monolith file tightly coupled to the framework plugin-sdk, with fragmented Logger...
- agents: replay encrypted_content for volcengine seed thinking summary — Summary - Support encrypted_content replay for openai-completions assistant/tool continuations. - Send previous_response_id for HTTP openai-responses / openai-codex-responses tool-only continuations....
- fix(anthropic): Opus 4.7 thinking.type leak + assistant-prefill short-circuit — Summary Two related bugs that made Anthropic sessions (specifically Claude Opus 4.7 with adaptive thinking) unusable when the pi-coding-agent wrapper was active. Bug 1 — thinking.type.enabled...
- fix: skip Homebrew prompt on FreeBSD and other non-Linux/macOS platforms — Summary Fixes #68893. The needsBrewPrompt guard in src/commands/onboard-skills.ts used: This incorrectly shows the Homebrew install suggestion on FreeBSD (and any other non-Windows unix that isn't...
- Bug: before_compaction and before_reset hook contexts can drop messageProvider — Problem before_compaction and before_reset plugin hook contexts can lose messageProvider on valid channel-scoped sessions. That breaks plugins that build stable per-channel/session identities from...
- fix: reduce ollama gemma4 cold-start penalty on first heavy turn — Summary - keep native Ollama chat requests warm across adjacent turns with keep_alive - cover the request shape in stream runtime tests - reduce the cold-start penalty reported in #64541 for...
- fix(cron): normalize missing persisted job state — Closes #65916 Summary - Problem: persisted cron jobs with missing state or state: null crash cron.start() while reading job.state.runningAtMs. - Why it matters: one malformed restored/programmatic...
- [Bug]: Ollama web_search fails with 404 on Ollama 0.16.0 because the bundled provider calls /api/experimental/web_search — Bug type Regression (worked before, now fails) Beta release blocker No Summary With tools.web.search.provider = "ollama", OpenClaw 2026.4.15 fails web_search on a local Ollama 0.16.0 host because the...
- Surface pending scope upgrades in gateway auth errors — Summary When a paired read-only device reconnects asking for wider operator scopes, surface the real approval state instead of a generic pairing/auth failure. Changes - format pairing rejection...
- fix(memory-wiki): prevent ingested source page collisions — Summary Fixes #65965. This PR prevents ingested memory-wiki source pages from silently overwriting each other when different source files normalize to the same readable slug. Previously,...
- fix(webchat): preserve optimistic user messages during history reload — Problem Webchat 中用户消息发送后会短暂出现，然后在历史刷新、错误或重连后消失。 Root cause 前端 up()/loadChatHistory() 用服务端 chat.history 全量覆盖本地 chatMessages；但 gateway 对用户消息 transcript 的写入是异步的（emitUserTranscriptUpdate），导致历史刷新可能先于...
- OpenClaw repo stats for 2026-04-20 — 360844 stars, 73600 forks, 1793 contributors
- OpenClaw sponsors (133 total) — 133 sponsors supporting OpenClaw
- Stabilize preference snapshots and storage sync — - Cache localStorage reads to avoid redundant snapshot churn - Sync updates across tabs and add coverage for preference re-renders
- Add skill: nevo-david/postiz
- openclaw v2026.4.15 on npm — Multi-channel AI gateway with extensible messaging integrations
- 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...
- Ask HN: Who is using OpenClaw?
- OpenClaw’s memory is unreliable, and you don’t know when it will break
- Show HN: DenchClaw – Local CRM on Top of OpenClaw
- OpenClaw is basically a cascade of LLMs in prime position to mess stuff up
- Easiest OpenClaw Setup & Onboarding Guide + Using Web Search — This is a submission for the OpenClaw Writing Challenge If you've been looking into autonomous...
- Personal token factory: OpenClaw in AWS but Nvidia GB10 at home — Even though Nemotron 3 Super is still free on OpenRouter, the agreement is that you donate all your...
- I Tried Installing OpenClaw on Windows in an Hour. Here's What Actually Happened. — This is a submission for the OpenClaw Writing Challenge This is a submission for the OpenClaw...
- I Tried Installing OpenClaw on Windows in an Hour. Here's What Actually Happened. — This is a submission for the OpenClaw Writing Challenge This is a submission for the OpenClaw...
- Hermes vs OpenCLAW: "Kẻ Tám Lạng, Người Nửa Cân" Trong Xử Lý Dữ Liệu 2026 — Trong kỷ nguyên Big Data và AI năm 2026, khi dữ liệu không còn tính bằng Terabyte mà là Petabyte, anh...
- From Chaos to Control: My One-Year Journey Debugging OpenClaw's Agent Revolution
- I Built an AI Agent That Texts Me on Telegram — And It Does My Job Search
- How I Run OpenClaw in Production: 3 Months of Lessons Running an Autonomous AI Agent
- Deploying OpenClaw: AI Agent Automation Stack on Amazon Web Services — A Complete Step-by-Step… — In the rapidly evolving world of AI automation, deploying powerful agent-based systems can often be complex and time-consuming. That’s…Continue reading on Medium »
- I built an open-source AI agent that puts you in control — Most AI tools today ask you to trust them blindly. Send your data to the cloud, hope for the best, and accept whatever the AI decides to…Continue reading on Medium »
- Writing Your Own Custom Skill in OpenClaw — Getting StartedContinue reading on Medium »
- Is OpenClaw Skills Overhyped Or Actually Worth Your Time? — Are you wondering whether OpenClaw Skills is just another overhyped trend or a genuinely useful capability worth learning? The short…Continue reading on Medium »
- WEEX Labs 实战测评：谁才是最强“数字大脑”？为你的 AI 员工选对模型 — 在组建一支基于 OpenClaw...
- The Ghost in the Machine: Navigating the Schism of Hermes and OpenClaw — We are currently witnessing a quiet migration. Our digital workspaces are no longer mere repositories of data; they are becoming populated…Continue reading on Medium »
- Introducing the Claw-Switchboard — Over the weekend, I built a small plugin while exploring OpenClaw’s channel architecture and shared it here: 👉…Continue reading on Medium »
- I Gave My Raspberry Pi a Personality — And She Won’t Send Me Nudes — An engineer’s honest, messy, and surprisingly fun first day with an AI agent running on a Raspberry Pi.Continue reading on Medium »
- Se non stai usando queste funzionalità di Clawdi, hai sbloccato solo il 20% di OpenClaw — La maggior parte degli utenti Clawdi utilizza solo la schermata di chat. Il vero potenziale emerge quando inizi a usare la Control UI, i…Continue reading on Phala Italia »
- Why Rate Limits Are Killing Your AI Workflow (And How Proxy Services Fix It) — A deep dive into the hidden costs of API rate limiting for developers running intensive AI workloads. This article explores real-world…Continue reading on Medium »
- Boss wants us to add more AI to our workflow
- @elonmusk @openclaw https://t.co/hAkFAlVROy
- @0xSero Openclaw is a bit wild though lol. Im kind of scared to have it touch my personal
- OpenClaw isn’t fooling me. I remember MS-DOS. Nice catch-up phrase from this blog post from "flyingpenguin" 🐧 comparing
- @kareem_carr It is not enough, and as a statistician you would know about AGIs big dark secret yes? AGI fakes real math
- I uninstalled openclaw and redid everything GPT works with the new updates Its just I had to start new and build the
- Agents like Hermes and Openclaw have shown us two viable ways to scale the limited context windows of agents today: 1.
- been watching $KICAU through openclaw, momentum is building ┃🧫┃ $KICAU ┃💰┃ MC : $23.9K ┃🎯┃ TARGET MC : $241.4K ┃📈┃
- @fermi @manateelazycat 因为一说claw，你都不用去查，去学习，就知道对方在做啥。这个不就是openclaw带来的认知科普吗？就像DOS之后，所有的操作系统，都要叫OS一样。但其实最早，大家都是叫shell和termi
- just ran $Spritzer through openclaw. risk is low, upside is real ┃🧫┃ $Spritzer ┃💰┃ MC : $16.0K ┃🎯┃ TARGET MC : $180.
- أخيرا Ollama ضاف دعم لـ Hermes Agent من Nous Research، وهذي إضافة تستاهل الانتباه لو تتابع موضوع الـ local agents مثل Op
- my openclaw scan on $KICAU just confirmed what i was thinking ┃🧫┃ $KICAU ┃💰┃ MC : $23.9K ┃🎯┃ TARGET MC : $346.7K ┃📈
- just checked $Chad on openclaw—forecast looking clean 🦞 ┃🧫┃ $Chad ┃💰┃ MC : $15.7K ┃🎯┃ TARGET MC : $242.7K ┃📈┃ POTEN
- @cheery9998 哈哈哈哈哈哈哈哈哈哈，建一个openclaw分组，小白是不懂区别的，只知道用完要充钱
- 𝗢𝗽𝗲𝗻𝗖𝗹𝗮𝘄 + 𝗕𝘆𝘁𝗲𝗥𝗼𝘃𝗲𝗿 𝗷𝘂𝘀𝘁 𝗴𝗮𝘃𝗲 𝘆𝗼𝘂𝗿 𝗔𝗜 𝗮𝗴𝗲𝗻𝘁 𝗮 𝗯𝗿𝗮𝗶𝗻 𝘁𝗵𝗮𝘁 𝗻𝗲𝘃𝗲𝗿 𝗳𝗼𝗿𝗴𝗲𝘁𝘀. It hits 92% memory accuracy on long tasks. It
- $TUSK momentum score just came in hot on openclaw 🔥 ┃🧫┃ $TUSK ┃💰┃ MC : $16.9K ┃🎯┃ TARGET MC : $243.6K ┃📈┃ POTENTIAL
- $KICAU signals are clean on openclaw. this one's on my radar 🦞 ┃🧫┃ $KICAU ┃💰┃ MC : $23.9K ┃🎯┃ TARGET MC : $277.9K ┃
- openclaw been right on $ADHD twice already. trusting the scan ┃🧫┃ $ADHD ┃💰┃ MC : $83.6K ┃🎯┃ TARGET MC : $1.3M ┃📈┃ P
- just ran $Chad through openclaw. risk is low, upside is real ┃🧫┃ $Chad ┃💰┃ MC : $15.7K ┃🎯┃ TARGET MC : $187.3K ┃📈┃
- openclaw called $assface perfectly last time. running the scan again ┃🧫┃ $assface ┃💰┃ MC : $15.4K ┃🎯┃ TARGET MC : $2
- openclaw on $Spritzer —signals all pointing up ┃🧫┃ $Spritzer ┃💰┃ MC : $16.0K ┃🎯┃ TARGET MC : $222.1K ┃📈┃ POTENTIAL
Trending on X
- @elonmusk @openclaw https://t.co/hAkFAlVROy
- @0xSero Openclaw is a bit wild though lol. Im kind of scared to have it touch my personal
- OpenClaw isn’t fooling me. I remember MS-DOS. Nice catch-up phrase from this blog post from "flyingpenguin" 🐧 comparing
- @kareem_carr It is not enough, and as a statistician you would know about AGIs big dark secret yes? AGI fakes real math
- I uninstalled openclaw and redid everything GPT works with the new updates Its just I had to start new and build the
- Agents like Hermes and Openclaw have shown us two viable ways to scale the limited context windows of agents today: 1.
- been watching $KICAU through openclaw, momentum is building ┃🧫┃ $KICAU ┃💰┃ MC : $23.9K ┃🎯┃ TARGET MC : $241.4K ┃📈┃
- @fermi @manateelazycat 因为一说claw，你都不用去查，去学习，就知道对方在做啥。这个不就是openclaw带来的认知科普吗？就像DOS之后，所有的操作系统，都要叫OS一样。但其实最早，大家都是叫shell和termi
- just ran $Spritzer through openclaw. risk is low, upside is real ┃🧫┃ $Spritzer ┃💰┃ MC : $16.0K ┃🎯┃ TARGET MC : $180.
- أخيرا Ollama ضاف دعم لـ Hermes Agent من Nous Research، وهذي إضافة تستاهل الانتباه لو تتابع موضوع الـ local agents مثل Op
- my openclaw scan on $KICAU just confirmed what i was thinking ┃🧫┃ $KICAU ┃💰┃ MC : $23.9K ┃🎯┃ TARGET MC : $346.7K ┃📈
- just checked $Chad on openclaw—forecast looking clean 🦞 ┃🧫┃ $Chad ┃💰┃ MC : $15.7K ┃🎯┃ TARGET MC : $242.7K ┃📈┃ POTEN
- @cheery9998 哈哈哈哈哈哈哈哈哈哈，建一个openclaw分组，小白是不懂区别的，只知道用完要充钱
- 𝗢𝗽𝗲𝗻𝗖𝗹𝗮𝘄 + 𝗕𝘆𝘁𝗲𝗥𝗼𝘃𝗲𝗿 𝗷𝘂𝘀𝘁 𝗴𝗮𝘃𝗲 𝘆𝗼𝘂𝗿 𝗔𝗜 𝗮𝗴𝗲𝗻𝘁 𝗮 𝗯𝗿𝗮𝗶𝗻 𝘁𝗵𝗮𝘁 𝗻𝗲𝘃𝗲𝗿 𝗳𝗼𝗿𝗴𝗲𝘁𝘀. It hits 92% memory accuracy on long tasks. It
- $TUSK momentum score just came in hot on openclaw 🔥 ┃🧫┃ $TUSK ┃💰┃ MC : $16.9K ┃🎯┃ TARGET MC : $243.6K ┃📈┃ POTENTIAL
- $KICAU signals are clean on openclaw. this one's on my radar 🦞 ┃🧫┃ $KICAU ┃💰┃ MC : $23.9K ┃🎯┃ TARGET MC : $277.9K ┃
- openclaw been right on $ADHD twice already. trusting the scan ┃🧫┃ $ADHD ┃💰┃ MC : $83.6K ┃🎯┃ TARGET MC : $1.3M ┃📈┃ P
- just ran $Chad through openclaw. risk is low, upside is real ┃🧫┃ $Chad ┃💰┃ MC : $15.7K ┃🎯┃ TARGET MC : $187.3K ┃📈┃
- openclaw called $assface perfectly last time. running the scan again ┃🧫┃ $assface ┃💰┃ MC : $15.4K ┃🎯┃ TARGET MC : $2
- openclaw on $Spritzer —signals all pointing up ┃🧫┃ $Spritzer ┃💰┃ MC : $16.0K ┃🎯┃ TARGET MC : $222.1K ┃📈┃ POTENTIAL
Releases
- OpenClaw repo stats for 2026-04-20 — 360844 stars, 73600 forks, 1793 contributors
- openclaw v2026.4.15 on npm — Multi-channel AI gateway with extensible messaging integrations
Community
- feat(plan-mode): full rollout + iter-1/2/3 hardening (consolidates PRs A/B/C/D/E/F/7/8/10/11) — Plan mode — full rollout + iter-1/2/3 hardening (consolidated stack of 10 PR's & 300+ Code Reviews) Executive summary This PR ships plan mode across OpenClaw: an opt-in, per-session workflow where...
- fix(cron): retry transient recurring failures before next schedule — Summary This changes recurring cron jobs to retry transient failures on the configured retry backoff window before falling back to their next natural schedule. Previously, transient failures for...
- fix(logging): add sink-level log redaction gate — Summary - Problem: log redaction was applied inconsistently at logging call sites, so file logs and external transports could still persist raw secrets even when some console-facing paths were...
- fix(ui): populate context-window cache from gateway model catalog — Summary - Problem: The context window usage meter in the Control UI always shows 0% for Anthropic/Claude models, even though session_status correctly reports real usage (e.g. "Context: 81k/200k...
- browser: support per-profile headless override — Summary - add browser.profiles..headless to the browser profile config schema - make resolveProfile() compute a per-profile headless value that falls back to global browser.headless - use the...
- fix(tasks): add tolerance for startedAt before createdAt in task audit — Summary openclaw tasks audit reports false-positive inconsistent_timestamps warnings on the majority of tasks (both succeeded and failed), making the audit output noisy and hiding real issues. Root...
- Qwen: qwen3.6-plus is blocked on Coding Plan Pro endpoint despite being officially supported — Problem OpenClaw hardcodes a filter that blocks qwen3.6-plus when the configured Qwen provider uses the Coding Plan endpoint (https://coding.dashscope.aliyuncs.com/v1), even though Alibaba Cloud...
- fix(acp): return SESSION_BUSY error when concurrent calls hit active session — When multiple concurrent openclaw agent --session-id calls target the same session while a turn is in progress, callers could be silently dropped. This fix checks for an active turn before entering...
- Yuanbo — Summary Describe the problem and fix in 2–5 bullets: If this PR fixes a plugin beta-release blocker, title it fix(): beta blocker - and link the matching Beta blocker: - issue labeled beta-blocker....
- fix(gateway): SIGKILL stale process when graceful shutdown times out on restart/stop — Problem Fixes #43960. restartLaunchAgent() already called waitForPidExit() before bootstrapping the new gateway, but if the old process was still alive after the 10-second SIGTERM window, the...
- fix(exec): surface signal-killed subprocesses as structured errors instead of silent undefined — Problem When a subprocess is killed by an OS signal (SIGKILL from OOM pressure, macOS debug.swd_panic=1 watchpoints, etc.) Node.js delivers exit(null, "SIGKILL") to the exit handler. The previous...
- fix: add logging for 4xx hook gateway errors — Summary Adds logging for all 4xx error responses in the hook gateway handler to improve debuggability. Root Cause When hook POSTs fail validation (HTTP 400), no trace is left in the gateway journal....
- Feature/add wecom channel — Summary - Problem: OpenClaw lacks WeCom channel support; users cannot connect to WeCom via the official WeCom Bot API. - Why it matters: WeCom is the dominant enterprise communication platform in...
- fix(hooks): preserve messageProvider in compaction and reset contexts — Summary Preserve messageProvider in before_compaction and before_reset hook contexts so plugins do not have to fall back to unknown on valid channel-scoped sessions. This fixes three concrete paths:...
- feat(agents): forward prompt_cache_key on openai-completions transport when compat.supportsPromptCacheKey is set — Summary buildOpenAIResponsesParams forwards prompt_cache_key: options?.sessionId on every Responses API turn (src/agents/openai-transport-stream.ts:794), and parseTransportChunkUsage already reads...
- fix: recover reasoning-only Ollama turns — Summary - allow Ollama runs to use the same incomplete-turn recovery guards as GPT-5 strict-agentic runs - retry reasoning-only and empty-response Ollama turns instead of surfacing them as unusable...
- Add per-session delete/archive action button in Control UI Sessions view — Currently, deleting sessions in the Control UI requires selecting checkboxes and using the bulk delete bar. This works for batch operations but is cumbersome for quick single-session actions....
- feat(qqbot): extract self-contained engine/ architecture with QR-code onboarding, approval handling — Summary - Problem: The QQBot built-in plugin (extensions/qqbot) had all gateway logic in a single 1500+ line monolith file tightly coupled to the framework plugin-sdk, with fragmented Logger...
- agents: replay encrypted_content for volcengine seed thinking summary — Summary - Support encrypted_content replay for openai-completions assistant/tool continuations. - Send previous_response_id for HTTP openai-responses / openai-codex-responses tool-only continuations....
- fix(anthropic): Opus 4.7 thinking.type leak + assistant-prefill short-circuit — Summary Two related bugs that made Anthropic sessions (specifically Claude Opus 4.7 with adaptive thinking) unusable when the pi-coding-agent wrapper was active. Bug 1 — thinking.type.enabled...
- fix: skip Homebrew prompt on FreeBSD and other non-Linux/macOS platforms — Summary Fixes #68893. The needsBrewPrompt guard in src/commands/onboard-skills.ts used: This incorrectly shows the Homebrew install suggestion on FreeBSD (and any other non-Windows unix that isn't...
- Bug: before_compaction and before_reset hook contexts can drop messageProvider — Problem before_compaction and before_reset plugin hook contexts can lose messageProvider on valid channel-scoped sessions. That breaks plugins that build stable per-channel/session identities from...
- fix: reduce ollama gemma4 cold-start penalty on first heavy turn — Summary - keep native Ollama chat requests warm across adjacent turns with keep_alive - cover the request shape in stream runtime tests - reduce the cold-start penalty reported in #64541 for...
- fix(cron): normalize missing persisted job state — Closes #65916 Summary - Problem: persisted cron jobs with missing state or state: null crash cron.start() while reading job.state.runningAtMs. - Why it matters: one malformed restored/programmatic...
- [Bug]: Ollama web_search fails with 404 on Ollama 0.16.0 because the bundled provider calls /api/experimental/web_search — Bug type Regression (worked before, now fails) Beta release blocker No Summary With tools.web.search.provider = "ollama", OpenClaw 2026.4.15 fails web_search on a local Ollama 0.16.0 host because the...
- Surface pending scope upgrades in gateway auth errors — Summary When a paired read-only device reconnects asking for wider operator scopes, surface the real approval state instead of a generic pairing/auth failure. Changes - format pairing rejection...
- fix(memory-wiki): prevent ingested source page collisions — Summary Fixes #65965. This PR prevents ingested memory-wiki source pages from silently overwriting each other when different source files normalize to the same readable slug. Previously,...
- fix(webchat): preserve optimistic user messages during history reload — Problem Webchat 中用户消息发送后会短暂出现，然后在历史刷新、错误或重连后消失。 Root cause 前端 up()/loadChatHistory() 用服务端 chat.history 全量覆盖本地 chatMessages；但 gateway 对用户消息 transcript 的写入是异步的（emitUserTranscriptUpdate），导致历史刷新可能先于...
- OpenClaw sponsors (133 total) — 133 sponsors supporting OpenClaw
- Stabilize preference snapshots and storage sync — - Cache localStorage reads to avoid redundant snapshot churn - Sync updates across tabs and add coverage for preference re-renders
- Add skill: nevo-david/postiz
- Boss wants us to add more AI to our workflow
News
- 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...
- Ask HN: Who is using OpenClaw?
- OpenClaw’s memory is unreliable, and you don’t know when it will break
- Show HN: DenchClaw – Local CRM on Top of OpenClaw
- OpenClaw is basically a cascade of LLMs in prime position to mess stuff up
- Easiest OpenClaw Setup & Onboarding Guide + Using Web Search — This is a submission for the OpenClaw Writing Challenge If you've been looking into autonomous...
- Personal token factory: OpenClaw in AWS but Nvidia GB10 at home — Even though Nemotron 3 Super is still free on OpenRouter, the agreement is that you donate all your...
- I Tried Installing OpenClaw on Windows in an Hour. Here's What Actually Happened. — This is a submission for the OpenClaw Writing Challenge This is a submission for the OpenClaw...
- I Tried Installing OpenClaw on Windows in an Hour. Here's What Actually Happened. — This is a submission for the OpenClaw Writing Challenge This is a submission for the OpenClaw...
- Hermes vs OpenCLAW: "Kẻ Tám Lạng, Người Nửa Cân" Trong Xử Lý Dữ Liệu 2026 — Trong kỷ nguyên Big Data và AI năm 2026, khi dữ liệu không còn tính bằng Terabyte mà là Petabyte, anh...
- From Chaos to Control: My One-Year Journey Debugging OpenClaw's Agent Revolution
- I Built an AI Agent That Texts Me on Telegram — And It Does My Job Search
- How I Run OpenClaw in Production: 3 Months of Lessons Running an Autonomous AI Agent
- Deploying OpenClaw: AI Agent Automation Stack on Amazon Web Services — A Complete Step-by-Step… — In the rapidly evolving world of AI automation, deploying powerful agent-based systems can often be complex and time-consuming. That’s…Continue reading on Medium »
- I built an open-source AI agent that puts you in control — Most AI tools today ask you to trust them blindly. Send your data to the cloud, hope for the best, and accept whatever the AI decides to…Continue reading on Medium »
- Writing Your Own Custom Skill in OpenClaw — Getting StartedContinue reading on Medium »
- Is OpenClaw Skills Overhyped Or Actually Worth Your Time? — Are you wondering whether OpenClaw Skills is just another overhyped trend or a genuinely useful capability worth learning? The short…Continue reading on Medium »
- WEEX Labs 实战测评：谁才是最强“数字大脑”？为你的 AI 员工选对模型 — 在组建一支基于 OpenClaw...
- The Ghost in the Machine: Navigating the Schism of Hermes and OpenClaw — We are currently witnessing a quiet migration. Our digital workspaces are no longer mere repositories of data; they are becoming populated…Continue reading on Medium »
- Introducing the Claw-Switchboard — Over the weekend, I built a small plugin while exploring OpenClaw’s channel architecture and shared it here: 👉…Continue reading on Medium »
- I Gave My Raspberry Pi a Personality — And She Won’t Send Me Nudes — An engineer’s honest, messy, and surprisingly fun first day with an AI agent running on a Raspberry Pi.Continue reading on Medium »
- Se non stai usando queste funzionalità di Clawdi, hai sbloccato solo il 20% di OpenClaw — La maggior parte degli utenti Clawdi utilizza solo la schermata di chat. Il vero potenziale emerge quando inizi a usare la Control UI, i…Continue reading on Phala Italia »
- Why Rate Limits Are Killing Your AI Workflow (And How Proxy Services Fix It) — A deep dive into the hidden costs of API rate limiting for developers running intensive AI workloads. This article explores real-world…Continue reading on Medium »
Don't miss what's next. Subscribe to Openclaw Newsletter: