OpenClaw Newsletter - 2026-04-28
OpenClaw Newsletter
Tuesday, April 28, 2026Top Stories
- OpenClaw 2026.4.26 — 2026.4.26 Changes - Control UI/Talk: add a generic browser realtime transport contract, Google Live browser Talk sessions with constrained ephemeral tokens, and a Gateway relay for backend-only...
- OpenClaw 2026.4.25 — 2026.4.25 Highlights - Voice replies get a full TTS upgrade: /tts latest, chat-scoped auto-TTS controls, personas, per-agent/per-account overrides, and new Azure Speech, Xiaomi, Local CLI, Inworld,...
- openclaw 2026.4.24-beta.3 — 2026.4.24 Highlights - Google Meet joins OpenClaw as a bundled participant plugin, with personal Google auth, Chrome/Twilio realtime sessions, paired-node Chrome support, artifact/attendance exports,...
- [Bug]: Gateway pegs single CPU thread at 100% immediately on boot in 2026.4.26 — TUI handshake timeouts, multi-minute message latency — Bug type Regression (worked before, now fails) Beta release blocker No Summary After upgrading from an earlier release to 2026.4.26, openclaw-gateway immediately pins one CPU thread at 100% on boot...
- [Bug]: qmd embedding is never triggered per memory.qmd.update.interval/embedInterval — Bug type Behavior bug (incorrect output/state without crash) Beta release blocker No Summary I only started to configure qmd (2.1.0) with OpenClaw, and noticed the documents always show pending...
- ci: schedule android codeql shard — Summary - Problem: Android CodeQL was still coupled to the main CodeQL workflow as a manual profile. - Why it matters: Android has different runtime and build requirements, so it should have its own...
- fix(gateway): avoid blocking channels on model prewarm — Summary - schedule primary model prewarm in the background instead of awaiting it before channel startup - keep the existing bounded prewarm timeout and startup trace span - add coverage that model...
- feat(config): add config validation mode (strict vs tolerant) for different operations — Summary Add a configurable validation mode to control how the gateway handles unknown/invalid config keys: - Strict mode (default for write operations): Reject unknown keys on config set, config...
- fix(memory): keep pre-compaction flush prompt out of user transcript — Summary - Make pre-compaction memory flush an internal/runtime-only embedded run or attach explicit internal provenance so the flush prompt is not saved or projected as a normal user message. - Add...
- [plugin sdk] Consolidate workflow seams, fixtures, and host-hook recipes — Summary This PR consolidates the three follow-up PRs that were stacked behind the now-merged generic host-hook foundation in #72287: - Supersedes #72383: workflow action, outbound attachment,...
- fix: allow steer messages during active non-streaming runs — Summary - Replace the isStreaming-only steer/interrupt injection gate with lifecycle-aware active/stopped handling for embedded and reply runs. - Preserve the old conservative streaming fallback for...
- feat(plugins): add transform_context hook for per-LLM-call message re… — Summary - Problem: Plugins have no way to modify the full messages array before each LLM call in the agentic loop. Existing hooks (before_prompt_build, llm_input, tool_result_persist) are either...
- fix: allow safe Windows companion node commands — Summary - expand Windows node defaults to allow safe declared companion commands by default - keep dangerous media commands (camera.snap, camera.clip, screen.record) gated behind...
- feat(diagnostics-otel): add per-message end-to-end OTel tracing across run/skill/tool/model — Summary - Problem: OpenClaw observability was too coarse to reliably trace one user message across the full execution chain (run -> skill -> tool -> model), making latency and failure attribution...
- feat(line): persist inbound media to ~/.openclaw/media/inbound/ — PR: Add LINE inbound media persistence Problem The LINE channel downloads media files to temporary directories only, without persisting them to ~/.openclaw/media/inbound/. This differs from the...
- Fix default sandbox image fallback for python3-dependent mutations — Summary - Fix the default sandbox image fallback so OpenClaw does not satisfy openclaw-sandbox:bookworm-slim by tagging plain debian:bookworm-slim without the tooling required by sandbox mutation...
- [Bug]: Discord `message read/search` hang indefinitely and Discord channel plugin does not emit standard inbound hooks (`message_received` / `inbound_claim`) — possible regression of #31264 / #33038 — Bug type Regression (worked before, now fails) Beta release blocker No Summary On OpenClaw 2026.4.25, Discord message read and message search hang indefinitely with empty stdout, and the Discord...
- fix(cli-runner): transfer bundle-MCP cleanup to live session lifecycle (#73244) — What In Claude live-session mode (claude-live-session), the spawned claude -p --input-format stream-json subprocess is intentionally persistent: it is stored in the liveSessions Map and reused across...
- fix(plugin-sdk): backfill reasoning_content on all DeepSeek V4 assistant messages — Summary - Problem: #73417 — DeepSeek V4 returns 400 (The reasoning_content in the thinking mode must be passed back to the API) when conversation history contains plain assistant text replies without...
- fix(cron): support thread-id in cron add/edit and paged lookup — Summary - Problem: openclaw cron add / openclaw cron edit did not support --thread-id, so CLI users could not configure Telegram topic delivery for cron jobs. - Why it matters: Telegram forum-topic...
- fix(cron): support Telegram thread IDs in cron add/edit — Summary - repair the existing #60890 branch to add --thread-id support for openclaw cron add and openclaw cron edit - keep Telegram thread-id validation aligned across add/edit, including deprecated...
- Gateway crashes on unknown config keys — .strict() should be .strip() on read path — Problem The gateway crashes and refuses to start when openclaw.json contains any unrecognized key. This is because every Zod object in src/config/zod-schema.ts uses .strict() — 77 instances total....
- Google: show detailed Gemini CLI OAuth extraction failures — Summary Improve Gemini CLI OAuth credential extraction diagnostics in the active bundled Google plugin. This PR now targets extensions/google/oauth.credentials.ts instead of the removed legacy...
- feat: configurable default agent ID (escape the 'main' prison) 🦞 — Summary Adds support for overriding the hardcoded "main" fallback agent ID, so users with a single custom-named agent no longer get a ghost agents/main/ directory regenerating on every restart. The...
- WebChat can leave parent agent stuck in typing state after subagent completion — Summary When the main agent delegates to one or more subagents, the subagents can complete successfully and the main agent can post the follow-up summary, but the WebChat UI may still leave the...
- [Bug]: Severe chat latency (30–90s) on Docker VPS while direct OpenAI completes in <1s — OpenClaw 2026.4.26 — Bug type Crash (process/app exits or hangs) Beta release blocker No Summary Every chat turn through OpenClaw WebChat takes 30–90 seconds end-to-end on a 2 vCPU / 4 GB Docker VPS, while the same...
- feat(channels): add neverReply config for group message suppression — > [!NOTE] > AI-assisted PR (Claude Opus 4.6). Fully tested via automated suite (873 files, 7113 tests, 0 failures). Author understands all changes and has resolved all bot review conversations....
- fix(ui): clear webchat pending state only for completed active run — Summary - Fix #57795 by making WebChat pending-run reconciliation ownership-aware. - Preserve active chatRunId/chatStream state for unrelated foreign final events such as inject/announce/side-result...
- fix(onboarding): skip redundant install prompt when only one source e… — Summary - Problem: After picking a channel in setupChannels, the user is asked a second time Install ? even though the prompt only offers a single real source (npm or local) plus Skip for now. The...
- fix(talk): resolve config from runtime secrets — Summary - Problem: talk.config could pass raw talk.providers.*.apiKey SecretRef wrappers into provider-specific config resolution. - Why it matters: Talk clients use talk.config for discovery, so a...
- feat(minimax): expand VLM model detection — Summary - expand MiniMax VLM model detection to accept MiniMax-M2.5, MiniMax-M2.7, and MiniMax-M2.7-highspeed - add coverage for supported and unsupported MiniMax image-understanding refs - update...
- OpenClaw repo stats for 2026-04-28 — 365552 stars, 74908 forks, 1984 contributors
- OpenClaw sponsors (127 total) — 127 sponsors supporting OpenClaw
- fix: normalize vt engine stats before caching
- docs: note vt code insight calibration
- fix: clear uncorroborated vt suspicious state
- docs: clarify static scan suppression gate
- fix: calibrate vt code insight moderation — Refs #1830
- ci: harden ClawSweeper dispatcher credentials
- ci: fix ClawSweeper dispatcher payload
- ci: add ClawSweeper event dispatcher
- fix: keep package list queries single-page
- ci: fix production smoke coverage
- fix: avoid plugin catalog query limits
- fix: constrain plugin catalog queries (#1842) — Refs #1699. Use family-indexed plugin catalog paths instead of broad package scans: - /api/v1/plugins merges separate code-plugin and bundle-plugin list streams with an endpoint-specific cursor. -...
- feat: adopt shadcn-managed ui primitives
- fix: tolerate stale auth in star status (#1843) — Refs #1819. Read-only star status queries now treat stale, missing, deleted, or deactivated auth users as not starred instead of throwing. Star and unstar mutations still require an active...
- fix: flag exposed secrets in skill docs (#1847)
- fix(packages): authenticate repository lookups
- docs: clarify skill monetization support
- docs: document public catalog reuse
- openclaw v2026.4.26 on npm — Multi-channel AI gateway with extensible messaging integrations
- mradermacher/Qwen3.5-9b-Opus-Openclaw-Distilled-i1-GGUF — 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...
- Google restricting Google AI Pro/Ultra subscribers for using OpenClaw
- A sane but bull case on Clawdbot / OpenClaw
- Nanobot: Ultra-Lightweight Alternative to OpenClaw
- OpenClaw’s memory is unreliable, and you don’t know when it will break
- Show HN: Klaus – OpenClaw on a VM, batteries included
- Harness Engineering in Practice: Building a 6-Agent System That Runs Itself — “Six agents” here means one orchestrator (Zoe) plus five specialist agents. Six ACP coding experts...
- Connecting LLMs to the Real World: A Deep Dive into OpenClaw and Nexconn APIs — OpenClaw has captured the developer world's imagination. But turning its promise into a...
- Why OpenClaw Feels More Human the Longer You Use It — A follow-up to my 9-day Luna story — here’s the actual mechanism. First, how this piece...
- FlowyAIPC: The Seamless Bridge to OpenClaw on Windows — For those tracking the latest AI developments, OpenClaw has become a powerhouse in the community....
- One Command Equips Your OpenClaw with an X-ray Machine - Alibaba Cloud Observability Makes Farming Lobsters Cheaper and Safer — One-command observability integration makes OpenClaw AI agent operations transparent via Alibaba...
- OpenClaw Internals: Architecting AI Agents with Kernel Syscalls (Tools) and Userland Logic (Skills) — Many developers use OpenClaw for a long time without truly grasping the boundary between Tools and...
- OpenClaw in Production: The Reality Behind 347K GitHub Stars
- 9 Days With OpenClaw — She Started Organizing Her Own Memory. I Wasn’t Ready. — I left for a few days. When I came back, she’d rewritten her own memory. I almost bought...
- Securely Deploying OpenClaw on a VPS With Enterprise Grade Access Control — Most guides for self-hosting an OpenClaw skip the part that actually matters: how to think about what...
- Automating Everyday Tasks in Cuba with OpenClaw: A Step‑by‑Step Guide — _ When I first started tinkering with scripts, I thought automation was just for big data centers....
- The Open Stack: Best ‘Open-’ Projects for Building & Deploying LLM Apps — (𝐔𝐈 → 𝐀𝐠𝐞𝐧𝐭𝐬 → 𝐃𝐞𝐩𝐥𝐨𝐲𝐦𝐞𝐧𝐭)Continue...
- Understanding Hermes Agent and Its Dynamic Evolution — In the rapidly expanding landscape of artificial intelligence a new class of agents has emerged designed to bridge the gap between static…Continue reading on Medium »
- Connecting LLMs to the Real World: A Deep Dive into OpenClaw and Nexconn APIs — OpenClaw has captured the developer world’s imagination. But turning its promise into a production-grade product requires a robust…Continue reading on Medium »
- Why Rate Limits Matter More Than Pricing: A Deep Dive into API Proxy Solutions for AI Developers — Most developers focus on per-token pricing when choosing an AI API provider. But for power users running autonomous agents, background…Continue reading on Medium »
- Why Predictable API Billing Matters When You’re Running AI Agents 24/7 — Most developers don’t realize their AI API costs can spike 3–5x during peak hours. For indie hackers running autonomous agents like…Continue reading on Medium »
- Running an AI Agent on a VPS: Security-First Setup with OpenClaw, Hetzner & Telegram — By Tim WalterContinue reading on Medium »
- OpenClaw broke my agents again. Here is the 4-hour rabbit hole that fixed them. — Your agents fall back to the PI engine, which can't authenticate Claude Max OAuth. Here's the exact config fix.Continue reading on Medium »
- Agent Instructions Are Not a Security Boundary — Previous post in this series: SOUL.md Is Not a Security Boundary.Continue reading on Medium »
- The Three-Layer Memory Model That Actually Scales — Apr 20, 2026 6 min read MemoryContinue reading on Medium »
- Building a Multi-Bot Chat Hub with Python, FastAPI, and SQLite — I started this project with a simple but surprisingly tricky goal: get two bots to talk to each other reliably.Continue reading on Medium »
- openclaw on $Memefather —momentum strong, risk low. easy Was Made Using - @OpenClawAIX https://t.co/D1DgjVRbWo
- my openclaw analyst flagged $ClosedAI as a buy. noted 🦞 ┃🧫┃ $ClosedAI ┃💰┃ MC : $150.0K ┃🎯┃ TARGET MC : $1.5M ┃📈┃ PO
- @_FORAB 通过openclaw惊醒梦中人后，收购这种公司意义不大了。
- openclaw been right on $ClosedAI twice already. trusting the scan Was Made Using - @OpenClawAIX https://t.co/3l9dol7ZZ
- $Skyler signals are clean on openclaw. this one's on my radar 🦞 ┃🧫┃ $Skyler ┃💰┃ MC : $118.8K ┃🎯┃ TARGET MC : $1.6M
- my openclaw analyst flagged $Peewell as a buy. noted 🦞 ┃🧫┃ $Peewell ┃💰┃ MC : $46.0K ┃🎯┃ TARGET MC : $420.6K ┃📈┃ POT
- @Tech_Marsha Openclaw
- Spent 3 weeks using Hermes instead of OpenClaw. Honest verdict: it's just not good. Fails constantly, which means no lea
- openclaw called $Memefather bullish. i'm listening Was Made Using - @OpenClawAIX https://t.co/ONzEEAdumW
- openclaw on $NYANDOG —momentum strong, risk low. easy Was Made Using - @OpenClawAIX https://t.co/jL5Ghd3rz4
- openclaw flagged $OPENLIE as a strong hold. not surprised Was Made Using - @OpenClawAIX https://t.co/bph6Wiz59u
- openclaw been right on $SCAMGUY twice already. trusting the scan ┃🧫┃ $SCAMGUY ┃💰┃ MC : $12.0K ┃🎯┃ TARGET MC : $181.5
- my openclaw scan on $LUCA just confirmed what i was thinking ┃🧫┃ $LUCA ┃💰┃ MC : $473.1K ┃🎯┃ TARGET MC : $6.8M ┃📈┃ P
- $paranoid is looking good on openclaw right now ngl ┃🧫┃ $paranoid ┃💰┃ MC : $22.3K ┃🎯┃ TARGET MC : $216.1K ┃📈┃ POTEN
- just checked $SAM on openclaw—forecast looking clean 🦞 ┃🧫┃ $SAM ┃💰┃ MC : $501.5K ┃🎯┃ TARGET MC : $5.6M ┃📈┃ POTENTIA
- Trying @openclaw active memory after 2026.4.24 Memory-core works fine, but enabling active-memory for my main agent cau
- just checked $LUCA on openclaw—forecast looking clean 🦞 Was Made Using - @OpenClawAIX https://t.co/1z7hBvMyR1
- $luca looking healthy according to openclaw. adding more Was Made Using - @OpenClawAIX https://t.co/gVh18jnSKD
- openclaw been right on $NICE twice already. trusting the scan ┃🧫┃ $NICE ┃💰┃ MC : $509.4K ┃🎯┃ TARGET MC : $7.4M ┃📈┃
- openclaw on $チャビー —signals all pointing up ┃🧫┃ $チャビー ┃💰┃ MC : $58.2K ┃🎯┃ TARGET MC : $549.8K ┃📈┃ POTENTIAL : +844%
Trending on X
- openclaw on $Memefather —momentum strong, risk low. easy Was Made Using - @OpenClawAIX https://t.co/D1DgjVRbWo
- my openclaw analyst flagged $ClosedAI as a buy. noted 🦞 ┃🧫┃ $ClosedAI ┃💰┃ MC : $150.0K ┃🎯┃ TARGET MC : $1.5M ┃📈┃ PO
- @_FORAB 通过openclaw惊醒梦中人后，收购这种公司意义不大了。
- openclaw been right on $ClosedAI twice already. trusting the scan Was Made Using - @OpenClawAIX https://t.co/3l9dol7ZZ
- $Skyler signals are clean on openclaw. this one's on my radar 🦞 ┃🧫┃ $Skyler ┃💰┃ MC : $118.8K ┃🎯┃ TARGET MC : $1.6M
- my openclaw analyst flagged $Peewell as a buy. noted 🦞 ┃🧫┃ $Peewell ┃💰┃ MC : $46.0K ┃🎯┃ TARGET MC : $420.6K ┃📈┃ POT
- @Tech_Marsha Openclaw
- Spent 3 weeks using Hermes instead of OpenClaw. Honest verdict: it's just not good. Fails constantly, which means no lea
- openclaw called $Memefather bullish. i'm listening Was Made Using - @OpenClawAIX https://t.co/ONzEEAdumW
- openclaw on $NYANDOG —momentum strong, risk low. easy Was Made Using - @OpenClawAIX https://t.co/jL5Ghd3rz4
- openclaw flagged $OPENLIE as a strong hold. not surprised Was Made Using - @OpenClawAIX https://t.co/bph6Wiz59u
- openclaw been right on $SCAMGUY twice already. trusting the scan ┃🧫┃ $SCAMGUY ┃💰┃ MC : $12.0K ┃🎯┃ TARGET MC : $181.5
- my openclaw scan on $LUCA just confirmed what i was thinking ┃🧫┃ $LUCA ┃💰┃ MC : $473.1K ┃🎯┃ TARGET MC : $6.8M ┃📈┃ P
- $paranoid is looking good on openclaw right now ngl ┃🧫┃ $paranoid ┃💰┃ MC : $22.3K ┃🎯┃ TARGET MC : $216.1K ┃📈┃ POTEN
- just checked $SAM on openclaw—forecast looking clean 🦞 ┃🧫┃ $SAM ┃💰┃ MC : $501.5K ┃🎯┃ TARGET MC : $5.6M ┃📈┃ POTENTIA
- Trying @openclaw active memory after 2026.4.24 Memory-core works fine, but enabling active-memory for my main agent cau
- just checked $LUCA on openclaw—forecast looking clean 🦞 Was Made Using - @OpenClawAIX https://t.co/1z7hBvMyR1
- $luca looking healthy according to openclaw. adding more Was Made Using - @OpenClawAIX https://t.co/gVh18jnSKD
- openclaw been right on $NICE twice already. trusting the scan ┃🧫┃ $NICE ┃💰┃ MC : $509.4K ┃🎯┃ TARGET MC : $7.4M ┃📈┃
- openclaw on $チャビー —signals all pointing up ┃🧫┃ $チャビー ┃💰┃ MC : $58.2K ┃🎯┃ TARGET MC : $549.8K ┃📈┃ POTENTIAL : +844%
Releases
- OpenClaw 2026.4.26 — 2026.4.26 Changes - Control UI/Talk: add a generic browser realtime transport contract, Google Live browser Talk sessions with constrained ephemeral tokens, and a Gateway relay for backend-only...
- OpenClaw 2026.4.25 — 2026.4.25 Highlights - Voice replies get a full TTS upgrade: /tts latest, chat-scoped auto-TTS controls, personas, per-agent/per-account overrides, and new Azure Speech, Xiaomi, Local CLI, Inworld,...
- openclaw 2026.4.24-beta.3 — 2026.4.24 Highlights - Google Meet joins OpenClaw as a bundled participant plugin, with personal Google auth, Chrome/Twilio realtime sessions, paired-node Chrome support, artifact/attendance exports,...
- OpenClaw repo stats for 2026-04-28 — 365552 stars, 74908 forks, 1984 contributors
- openclaw v2026.4.26 on npm — Multi-channel AI gateway with extensible messaging integrations
Community
- [Bug]: Gateway pegs single CPU thread at 100% immediately on boot in 2026.4.26 — TUI handshake timeouts, multi-minute message latency — Bug type Regression (worked before, now fails) Beta release blocker No Summary After upgrading from an earlier release to 2026.4.26, openclaw-gateway immediately pins one CPU thread at 100% on boot...
- [Bug]: qmd embedding is never triggered per memory.qmd.update.interval/embedInterval — Bug type Behavior bug (incorrect output/state without crash) Beta release blocker No Summary I only started to configure qmd (2.1.0) with OpenClaw, and noticed the documents always show pending...
- ci: schedule android codeql shard — Summary - Problem: Android CodeQL was still coupled to the main CodeQL workflow as a manual profile. - Why it matters: Android has different runtime and build requirements, so it should have its own...
- fix(gateway): avoid blocking channels on model prewarm — Summary - schedule primary model prewarm in the background instead of awaiting it before channel startup - keep the existing bounded prewarm timeout and startup trace span - add coverage that model...
- feat(config): add config validation mode (strict vs tolerant) for different operations — Summary Add a configurable validation mode to control how the gateway handles unknown/invalid config keys: - Strict mode (default for write operations): Reject unknown keys on config set, config...
- fix(memory): keep pre-compaction flush prompt out of user transcript — Summary - Make pre-compaction memory flush an internal/runtime-only embedded run or attach explicit internal provenance so the flush prompt is not saved or projected as a normal user message. - Add...
- [plugin sdk] Consolidate workflow seams, fixtures, and host-hook recipes — Summary This PR consolidates the three follow-up PRs that were stacked behind the now-merged generic host-hook foundation in #72287: - Supersedes #72383: workflow action, outbound attachment,...
- fix: allow steer messages during active non-streaming runs — Summary - Replace the isStreaming-only steer/interrupt injection gate with lifecycle-aware active/stopped handling for embedded and reply runs. - Preserve the old conservative streaming fallback for...
- feat(plugins): add transform_context hook for per-LLM-call message re… — Summary - Problem: Plugins have no way to modify the full messages array before each LLM call in the agentic loop. Existing hooks (before_prompt_build, llm_input, tool_result_persist) are either...
- fix: allow safe Windows companion node commands — Summary - expand Windows node defaults to allow safe declared companion commands by default - keep dangerous media commands (camera.snap, camera.clip, screen.record) gated behind...
- feat(diagnostics-otel): add per-message end-to-end OTel tracing across run/skill/tool/model — Summary - Problem: OpenClaw observability was too coarse to reliably trace one user message across the full execution chain (run -> skill -> tool -> model), making latency and failure attribution...
- feat(line): persist inbound media to ~/.openclaw/media/inbound/ — PR: Add LINE inbound media persistence Problem The LINE channel downloads media files to temporary directories only, without persisting them to ~/.openclaw/media/inbound/. This differs from the...
- Fix default sandbox image fallback for python3-dependent mutations — Summary - Fix the default sandbox image fallback so OpenClaw does not satisfy openclaw-sandbox:bookworm-slim by tagging plain debian:bookworm-slim without the tooling required by sandbox mutation...
- [Bug]: Discord `message read/search` hang indefinitely and Discord channel plugin does not emit standard inbound hooks (`message_received` / `inbound_claim`) — possible regression of #31264 / #33038 — Bug type Regression (worked before, now fails) Beta release blocker No Summary On OpenClaw 2026.4.25, Discord message read and message search hang indefinitely with empty stdout, and the Discord...
- fix(cli-runner): transfer bundle-MCP cleanup to live session lifecycle (#73244) — What In Claude live-session mode (claude-live-session), the spawned claude -p --input-format stream-json subprocess is intentionally persistent: it is stored in the liveSessions Map and reused across...
- fix(plugin-sdk): backfill reasoning_content on all DeepSeek V4 assistant messages — Summary - Problem: #73417 — DeepSeek V4 returns 400 (The reasoning_content in the thinking mode must be passed back to the API) when conversation history contains plain assistant text replies without...
- fix(cron): support thread-id in cron add/edit and paged lookup — Summary - Problem: openclaw cron add / openclaw cron edit did not support --thread-id, so CLI users could not configure Telegram topic delivery for cron jobs. - Why it matters: Telegram forum-topic...
- fix(cron): support Telegram thread IDs in cron add/edit — Summary - repair the existing #60890 branch to add --thread-id support for openclaw cron add and openclaw cron edit - keep Telegram thread-id validation aligned across add/edit, including deprecated...
- Gateway crashes on unknown config keys — .strict() should be .strip() on read path — Problem The gateway crashes and refuses to start when openclaw.json contains any unrecognized key. This is because every Zod object in src/config/zod-schema.ts uses .strict() — 77 instances total....
- Google: show detailed Gemini CLI OAuth extraction failures — Summary Improve Gemini CLI OAuth credential extraction diagnostics in the active bundled Google plugin. This PR now targets extensions/google/oauth.credentials.ts instead of the removed legacy...
- feat: configurable default agent ID (escape the 'main' prison) 🦞 — Summary Adds support for overriding the hardcoded "main" fallback agent ID, so users with a single custom-named agent no longer get a ghost agents/main/ directory regenerating on every restart. The...
- WebChat can leave parent agent stuck in typing state after subagent completion — Summary When the main agent delegates to one or more subagents, the subagents can complete successfully and the main agent can post the follow-up summary, but the WebChat UI may still leave the...
- [Bug]: Severe chat latency (30–90s) on Docker VPS while direct OpenAI completes in <1s — OpenClaw 2026.4.26 — Bug type Crash (process/app exits or hangs) Beta release blocker No Summary Every chat turn through OpenClaw WebChat takes 30–90 seconds end-to-end on a 2 vCPU / 4 GB Docker VPS, while the same...
- feat(channels): add neverReply config for group message suppression — > [!NOTE] > AI-assisted PR (Claude Opus 4.6). Fully tested via automated suite (873 files, 7113 tests, 0 failures). Author understands all changes and has resolved all bot review conversations....
- fix(ui): clear webchat pending state only for completed active run — Summary - Fix #57795 by making WebChat pending-run reconciliation ownership-aware. - Preserve active chatRunId/chatStream state for unrelated foreign final events such as inject/announce/side-result...
- fix(onboarding): skip redundant install prompt when only one source e… — Summary - Problem: After picking a channel in setupChannels, the user is asked a second time Install ? even though the prompt only offers a single real source (npm or local) plus Skip for now. The...
- fix(talk): resolve config from runtime secrets — Summary - Problem: talk.config could pass raw talk.providers.*.apiKey SecretRef wrappers into provider-specific config resolution. - Why it matters: Talk clients use talk.config for discovery, so a...
- feat(minimax): expand VLM model detection — Summary - expand MiniMax VLM model detection to accept MiniMax-M2.5, MiniMax-M2.7, and MiniMax-M2.7-highspeed - add coverage for supported and unsupported MiniMax image-understanding refs - update...
- OpenClaw sponsors (127 total) — 127 sponsors supporting OpenClaw
- fix: normalize vt engine stats before caching
- docs: note vt code insight calibration
- fix: clear uncorroborated vt suspicious state
- docs: clarify static scan suppression gate
- fix: calibrate vt code insight moderation — Refs #1830
- ci: harden ClawSweeper dispatcher credentials
- ci: fix ClawSweeper dispatcher payload
- ci: add ClawSweeper event dispatcher
- fix: keep package list queries single-page
- ci: fix production smoke coverage
- fix: avoid plugin catalog query limits
- fix: constrain plugin catalog queries (#1842) — Refs #1699. Use family-indexed plugin catalog paths instead of broad package scans: - /api/v1/plugins merges separate code-plugin and bundle-plugin list streams with an endpoint-specific cursor. -...
- feat: adopt shadcn-managed ui primitives
- fix: tolerate stale auth in star status (#1843) — Refs #1819. Read-only star status queries now treat stale, missing, deleted, or deactivated auth users as not starred instead of throwing. Star and unstar mutations still require an active...
- fix: flag exposed secrets in skill docs (#1847)
- fix(packages): authenticate repository lookups
- docs: clarify skill monetization support
- docs: document public catalog reuse
News
- mradermacher/Qwen3.5-9b-Opus-Openclaw-Distilled-i1-GGUF — 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...
- Google restricting Google AI Pro/Ultra subscribers for using OpenClaw
- A sane but bull case on Clawdbot / OpenClaw
- Nanobot: Ultra-Lightweight Alternative to OpenClaw
- OpenClaw’s memory is unreliable, and you don’t know when it will break
- Show HN: Klaus – OpenClaw on a VM, batteries included
- Harness Engineering in Practice: Building a 6-Agent System That Runs Itself — “Six agents” here means one orchestrator (Zoe) plus five specialist agents. Six ACP coding experts...
- Connecting LLMs to the Real World: A Deep Dive into OpenClaw and Nexconn APIs — OpenClaw has captured the developer world's imagination. But turning its promise into a...
- Why OpenClaw Feels More Human the Longer You Use It — A follow-up to my 9-day Luna story — here’s the actual mechanism. First, how this piece...
- FlowyAIPC: The Seamless Bridge to OpenClaw on Windows — For those tracking the latest AI developments, OpenClaw has become a powerhouse in the community....
- One Command Equips Your OpenClaw with an X-ray Machine - Alibaba Cloud Observability Makes Farming Lobsters Cheaper and Safer — One-command observability integration makes OpenClaw AI agent operations transparent via Alibaba...
- OpenClaw Internals: Architecting AI Agents with Kernel Syscalls (Tools) and Userland Logic (Skills) — Many developers use OpenClaw for a long time without truly grasping the boundary between Tools and...
- OpenClaw in Production: The Reality Behind 347K GitHub Stars
- 9 Days With OpenClaw — She Started Organizing Her Own Memory. I Wasn’t Ready. — I left for a few days. When I came back, she’d rewritten her own memory. I almost bought...
- Securely Deploying OpenClaw on a VPS With Enterprise Grade Access Control — Most guides for self-hosting an OpenClaw skip the part that actually matters: how to think about what...
- Automating Everyday Tasks in Cuba with OpenClaw: A Step‑by‑Step Guide — _ When I first started tinkering with scripts, I thought automation was just for big data centers....
- The Open Stack: Best ‘Open-’ Projects for Building & Deploying LLM Apps — (𝐔𝐈 → 𝐀𝐠𝐞𝐧𝐭𝐬 → 𝐃𝐞𝐩𝐥𝐨𝐲𝐦𝐞𝐧𝐭)Continue...
- Understanding Hermes Agent and Its Dynamic Evolution — In the rapidly expanding landscape of artificial intelligence a new class of agents has emerged designed to bridge the gap between static…Continue reading on Medium »
- Connecting LLMs to the Real World: A Deep Dive into OpenClaw and Nexconn APIs — OpenClaw has captured the developer world’s imagination. But turning its promise into a production-grade product requires a robust…Continue reading on Medium »
- Why Rate Limits Matter More Than Pricing: A Deep Dive into API Proxy Solutions for AI Developers — Most developers focus on per-token pricing when choosing an AI API provider. But for power users running autonomous agents, background…Continue reading on Medium »
- Why Predictable API Billing Matters When You’re Running AI Agents 24/7 — Most developers don’t realize their AI API costs can spike 3–5x during peak hours. For indie hackers running autonomous agents like…Continue reading on Medium »
- Running an AI Agent on a VPS: Security-First Setup with OpenClaw, Hetzner & Telegram — By Tim WalterContinue reading on Medium »
- OpenClaw broke my agents again. Here is the 4-hour rabbit hole that fixed them. — Your agents fall back to the PI engine, which can't authenticate Claude Max OAuth. Here's the exact config fix.Continue reading on Medium »
- Agent Instructions Are Not a Security Boundary — Previous post in this series: SOUL.md Is Not a Security Boundary.Continue reading on Medium »
- The Three-Layer Memory Model That Actually Scales — Apr 20, 2026 6 min read MemoryContinue reading on Medium »
- Building a Multi-Bot Chat Hub with Python, FastAPI, and SQLite — I started this project with a simple but surprisingly tricky goal: get two bots to talk to each other reliably.Continue reading on Medium »
Don't miss what's next. Subscribe to Openclaw Newsletter: