OpenClaw Newsletter - 2026-06-19
OpenClaw Newsletter
Friday, June 19, 2026Top Stories
- openclaw 2026.6.9-beta.1 — 2026.6.9 Highlights - Richer Telegram delivery: Telegram now sends rich HTML, preserves rich markdown and sticker paths, renders progress drafts and command output more faithfully, and keeps mentions...
- fix(usage): exclude untimestamped records from daily ranges — Summary What problem does this PR solve? - Fixes ranged dashboard/session usage summaries so untimestamped cached transcript rows do not inflate selected-day totals or model usage. - Keeps all-time...
- fix(tailscale): reset stale serve entries before adding new ones on startup — Fixes #93936 Summary tailscale serve is append-only and never cleans up old entries. Each gateway restart with gateway.tailscale.mode: "serve" adds a new serve entry without removing stale ones,...
- Fix Discord default account startup priority — Fixes #77429 Real behavior proof - Behavior or issue addressed: Discord multi-account startup delay ordering previously followed the configured account id list directly, so a configured primary...
- fix(heartbeat): suppress private text delivery when heartbeat_respond tool is available but unused — Summary When heartbeat runs in message_tool_only mode and the model is given the heartbeat_respond tool but does NOT call it, the raw text response was being delivered to the Telegram channel —...
- docs(install): add vendor-neutral NVIDIA Jetson install guide — Summary Add a vendor-neutral NVIDIA Jetson install guide (docs/install/jetson.md) parallel to the existing raspberry-pi.md. Jetson modules (Orin, Xavier) are increasingly common always-on ARM64 edge...
- [Bug]: Skill loader ignores configured workspace path and hardcodes fallback to ~/.openclaw/workspace — Bug type Behavior bug (incorrect output/state without crash) Beta release blocker No Summary When a custom workspace directory is configured (e.g., on drive E:\), OpenClaw still attempts to load...
- [Bug]: Codex app-server session wedges indefinitely after a 429 rate-limit burst — terminal-idle watchdog still gated by `activeAppServerTurnRequests > 0` (regression/incomplete fix of #82681) — Bug type Behavior bug (incorrect state without crash) — session wedge / message loss. Summary On 2026.5.28, a Codex app-server session can wedge indefinitely with the agent silently not replying —...
- docs: vendor-neutral "Run OpenClaw on NVIDIA Jetson / edge hardware" guide — and how to reference turnkey devices? — Summary docs/install/ covers running the Gateway on Raspberry Pi and on specific VPS/cloud vendors (Fly, Hetzner, GCP, Azure, exe.dev…), but there's no guide for NVIDIA Jetson / dedicated edge...
- fix(auto-reply): guard context report tool entries — Summary - Guard cached context report tool entries before /context detail and /context map read tool names and size fields. - Reuse one normalized tool-report view for text diagnostics and the...
- fix(diagnostic): lower stall abort threshold and extend model-call recovery to CLI sessions — Summary The diagnostic heartbeat detects stalled sessions (model calls stuck >120s) but the abort threshold was 360s (6 minutes) — far too long to recover from a hung provider. Additionally,...
- fix(agents): guard message provider tool name reads — Summary - Guard filterToolsByMessageProvider() against tool descriptors whose name getter throws while provider allow/deny policy is applied. - Avoid reading descriptor names when no message-provider...
- fix(agents): guard lean tool name reads — Summary - Catch-bound local-model lean tool name normalization so unreadable descriptors cannot crash lean filtering. - Drop malformed or empty-name descriptors only while lean filtering is active;...
- fix(agents): guard deferred follow-up tool descriptors — Summary - Guard deferred follow-up guidance setup against active tools whose name getter throws. - Preserve unreadable-name tools in the returned list while still updating healthy exec/process...
- [codex] feat(mcp): stage binary tool content for relay — Summary Part of #87023. This adds a conservative relay foundation for MCP binary content: - stage MCP image, audio, and blob-backed resource content into OpenClaw-owned outbound media attachments -...
- fix(agents): guard code mode catalog tool names — Summary - Guard Code Mode catalog filtering against tools whose name getter throws before the shared Tool Search catalog compaction path runs. - Keep marked Code Mode control tools visible, drop...
- test(cron): cover cron base-session preservation during reaper sweep — Summary Adds regression coverage for the current cron session reaper boundary: expired canonical agent::cron::run: run-session descendants are pruned, while stable agent::cron: base rows are...
- fix(plugin-sdk): guard plain text tool name reads — Summary - Guard plain-text tool-call compatibility against unreadable tool name accessors in provider stream context. - Skip only malformed/unreadable context tool entries while preserving healthy...
- fix(cron): prune isolated-target cron sessions in session reaper — Summary - isCronRunSessionKey only matches cron session keys containing :run: (cron:jobId:run:runId). Isolated-target cron jobs (sessionTarget: isolated) generate base session keys without :run:...
- fix(text): strip commentary to=functions.* trace lines from assistant output (fixes #89668) — What does this PR do? Strips Anthropic-style commentary to=functions.* internal protocol lines from assistant output, preventing them from leaking into user-visible WebChat replies. Related Issue...
- ACP subagent deliver fails: streamTo silent stall + subagent_announce "No scope" fallback errors in 5.28 — Summary Two failure modes in ACP subagent announce/fallback path cause 100% delivery failure when the main session is fast (5–15s/turn). Both stem from the same root: the announce path does not...
- fix(agents): avoid enumerating wrapped tool descriptors — Summary - Stop before-tool-call and abort wrappers from spreading arbitrary active tool descriptors. - Add a shared wrapper descriptor copier that preserves the AgentTool contract fields plus...
- fix(providers): strip cache-boundary marker from non-Anthropic prompts — Summary - The internal SYSTEM_PROMPT_CACHE_BOUNDARY marker (\n\n) is meant to stay private to OpenClaw's Anthropic prompt-cache split. It currently leaks verbatim into the system prompt sent to...
- fix(control-ui): clear stale composer replays after send — Summary - After sending a message, the Control UI composer clears the draft. However, a stale handleInput event can replay the just-submitted text back into the textarea, making it appear as if the...
- [Bug]: Dashboard "今日模型调用" shows historical cumulative data instead of daily usage after v2026.5.28 upgrade — Bug type Behavior bug (incorrect output/state without crash) Beta release blocker No Summary After upgrading from v2026.5.27 to v2026.5.28, the Dashboard "Usage" page ("今日模型调用" / today's model calls)...
- fix(gateway): allow macOS app platform version refresh without re-pairing — Summary - macOS app enters infinite "Approve this Mac app to control OpenClaw" dialog loop after an OS update (e.g. macOS 26.5.0 → 26.5.1) - Root cause: resolvePinnedClientMetadata treats any macOS...
- Remove legacy tools.exec.security config in favor of tools.exec.mode — Summary tools.exec.security should be removed from the steady-state config surface and normalized into tools.exec.mode. Existing configs that still use the legacy key should be migrated by openclaw...
- fix(agents): snapshot client tool descriptors — Summary - Snapshot hosted/client tool descriptors once per attempt before reserved-name checks, Tool Search planning, allowlist collection, and registration. - Skip unreadable or malformed client...
- OpenClaw repo stats for 2026-06-19 — 379454 stars, 79431 forks, 2614 contributors
- OpenClaw sponsors (107 total) — 107 sponsors supporting OpenClaw
- fix: show taxonomy on plugin cards (#2748) — Restore resolved plugin taxonomy on sorted catalog cards and show controlled category labels when author topics are absent. Prepared head SHA: eed82783f0ce23cc6744c0cde1bf2dd24dd1a804 Reviewed-by:...
- feat: surface plugin manifest icons (#2747)
- feat: add skill install backfill migration — Adds the aggregate skill install backfill migration, tests, and operator proof.
- feat: derive skill icons from categories (#2745)
- fix: persist registry backup seed completion state (#2746)
- ci: shard local-auth playwright gate (#2743)
- feat: enforce Convex retention policy — Add typed Convex retention policies, auth session/refresh-token cleanup crons, and retention guidance.
- feat: parallelize registry artifact backup backfill (#2741)
- feat: move topic discovery into search and category browse (#2732) — feat: search categories and topics fix: remove topic filters from browse sidebars feat: show top topics for selected categories fix(deps): resolve static audit advisories fix(search): scope skill...
- fix: stop auto-greeting staff emails (#2738)
- Add package runtime id repair command (#2734) — feat: add package runtime id repair chore: refresh audit gate overrides
- fix: shard skill stat event pruning
- openclaw-cli 2026.6.8 on Homebrew — Your own personal AI assistant
- sulpikar2/Qwen3.5-4B-Claude-4.6-Opus-Reasoning-openclaw
- Iter435/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...
- Tell HN: Anthropic no longer allowing Claude Code subscriptions to use OpenClaw
- Anthropic says OpenClaw-style Claude CLI usage is allowed again
- OpenClaw isn't fooling me. I remember MS-DOS
- AI Assistant Architecture: LLM, Memory, Tools, Routing, Observability — A production AI assistant is not "an LLM with a prompt". It is a system that accepts intent, keeps...
- How I connected OpenClaw to a Chinese frontier model and cut agent costs by 90% — I've been running OpenClaw for a few months. Love the agent workflow. Not so much the monthly...
- I think I found the first real reason to build AI agent workflows in OpenClaw — A practical case for using OpenClaw to automate receipt-to-ledger workflows: bounded inputs and outputs, clear human review gates, and a better split
- The journey to a healthcare AI personal assistant — Being unexpectedly immersed in the challenge of supporting a family member with a rare neurodegenerative condition is deeply distressing…Continue reading on Medium »
- Your AI Coding Assistant Should Still Be Working While You Sleep — The laptop was warm when I woke up.Continue reading on Medium »
- Building a Monthly Personal Review Agent with OpenClaw — Why I Built ItContinue reading on Medium »
- How to Use OpenClaw to Boost Your WordPress Workflow — Managing a WordPress website involves much more than publishing articles. You may need to research topics, write content, optimize for…Continue reading on Medium »
- Looking for an OpenClaw Alternative? Here Are 7 Worth Exploring in 2026 — AI agents have evolved far beyond simple chatbots. Tools like OpenClaw can automate tasks, remember context, interact with apps, and…Continue reading on Medium »
- Private GPT — Ask questions to your documents without an internet connection, using the power of LLMs. 100% private, no data leaves your execution environment at any point. You can ingest documents and ask...
- ChatSonic — Combining Google's knowledge graph, ChatSonic delivers real-time data, voice interaction, AI-produced visual content, personalized avatars, and API for services. It revisits past dialogs for informed...
- Reins — Empowering LLM researchers and hobbyists with seamless control over self-hosted models. Connect remotely, customize prompts, manage chats, and fine-tune configurations. All in one intuitive app.
- LlamaGPT — LlamaGPT is a chatbot that provides a ChatGPT-like experience, with no data leaving your device.
- RT @chuhaiqu: 来自中国的硬件团队的这个 PicoClaw 牛逼，把 OpenClaw（前Clawdbot）用 Go 彻底重构了。 直接把 AI Agent 的硬件门槛打到了地板，内存占用不到 10MB，启动仅需 1 秒。以前
- 3/ OpenClaw v2026.6.8 发布，近期 PR 全在做运行时治理： · tools.fs.roots：按 Agent 划定文件系统根目录与访问模式（权限收敛） · context-engine interceptCompac
- OpenClaw / Agent 日报 · 2026-06-19 今天主线很清晰：Agent 的工具调用与权限正在从「能用」走向「可治理」。 MCP 企业 OAuth、A2A 协议落地、AI Agent 被用于真实攻防——三件事指向同一 — OpenClaw / Agent 日报 · 2026-06-19 今天主线很清晰：Agent 的工具调用与权限正在从「能用」走向「可治理」。 MCP 企业 OAuth、A2A 协议落地、AI Agent 被用于真实攻防——三件事指向同一个结论：当 Agent 握住凭证和外部动作，安全与权限就是生产前提。🧵
- Your own AI Private and always on assistant! No setup required and you can sign with X! 🦞 @PaioClaw is a managed OpenC
- @teneo_protocol The setup guides across different IDEs and AI coding agents are a nice touch. Looking forward to seeing
- OpenClawﾁｬﾝ、いずれ Mac miniにお引越ししたいけど、ラズパイのうちはラズパイにしかできないこともあるなあ。ラズパイだとおでかけもできるのか
- C'est le 3è jour de #VivaTech 2026, et le 3è épisode de la Quotidienne by @LesVoixdelaTech avec @benjaminvincent vous at
- @jeremyphoward @Zai_org Is it outputting Chinese characters during long conversations? That was at least a problem for m
- RT @elonmusk: People giving OpenClaw root access to their entire life https://t.co/jqW0FwiW03
- DietrichGebert/ponytail pushed `openclaw-ponytail-skill-sync` - suggesting integration with the openclaw agent framework — DietrichGebert/ponytail pushed openclaw-ponytail-skill-sync - suggesting integration with the openclaw agent framework. The repo accumulated 800+ stars without visible release triggers in commit...
- RT @gladstein: Three months ago I wrote my first reflections on AI I had been using OpenClaw for a few weeks, and it bl
- RT @empiriolabsai: Always-on OpenClaw or Hermes agents on EmpirioLabs. Simple monthly pricing, billed to credits, cancel
- And that's the most common thing in innovation. Small players have larger risk potential and hence they adopt new techno
- @leumas4samuel @jayhemz “Do you know how outlier works?” When I’m literally on OpenClaw Atlas after getting done with Ae
- RT @BlockView0214: 除了OpenClaw/Hermes，GitHub 上这波开源个人 agent，已经比想象中成熟： CowAgent（45k+ stars）：开源超级 AI 助手，能规划任务、跑工具、维护记忆和知识，前
- This community is for: - programmers - Indiehackers - Founder - AI / openclaw builders - iOS app builders — This community is for: - programmers - Indiehackers - Founder - AI / openclaw builders - iOS app builders
- Quick recap from ClawCon Nairobi yesterday, the first ClawCon on the African continent. 910 people registered. 400 show
- RT @murasametech: Raspberry Pi でOpenClawを運用する方法を以下の書籍で解説しています👇 https://t.co/pU1J65pa3o
- RT @wanerfu: 🔥 一条 AI 短剧，狂揽 41.8 万赞！ AI 短剧赛道，已经进化到“工业流水线”时代 爆款工作流：OpenClaw + XCrawl + Seedance 2.0 把短剧制作门槛，直接打到了地板上 详 — RT @wanerfu: 🔥 一条 AI 短剧，狂揽 41.8 万赞！ AI 短剧赛道，已经进化到“工业流水线”时代 爆款工作流：OpenClaw + XCrawl + Seedance 2.0 把短剧制作门槛，直接打到了地板上 详细 Skill 配置 + 操作步骤，…
- À VivaTech, Thibault Sottiaux (OpenAI) et Peter Steinberger (OpenClaw) racontent le tournant de l’IA agentique https://t
Trending on X
- RT @chuhaiqu: 来自中国的硬件团队的这个 PicoClaw 牛逼，把 OpenClaw（前Clawdbot）用 Go 彻底重构了。 直接把 AI Agent 的硬件门槛打到了地板，内存占用不到 10MB，启动仅需 1 秒。以前
- 3/ OpenClaw v2026.6.8 发布，近期 PR 全在做运行时治理： · tools.fs.roots：按 Agent 划定文件系统根目录与访问模式（权限收敛） · context-engine interceptCompac
- OpenClaw / Agent 日报 · 2026-06-19 今天主线很清晰：Agent 的工具调用与权限正在从「能用」走向「可治理」。 MCP 企业 OAuth、A2A 协议落地、AI Agent 被用于真实攻防——三件事指向同一 — OpenClaw / Agent 日报 · 2026-06-19 今天主线很清晰：Agent 的工具调用与权限正在从「能用」走向「可治理」。 MCP 企业 OAuth、A2A 协议落地、AI Agent 被用于真实攻防——三件事指向同一个结论：当 Agent 握住凭证和外部动作，安全与权限就是生产前提。🧵
- Your own AI Private and always on assistant! No setup required and you can sign with X! 🦞 @PaioClaw is a managed OpenC
- @teneo_protocol The setup guides across different IDEs and AI coding agents are a nice touch. Looking forward to seeing
- OpenClawﾁｬﾝ、いずれ Mac miniにお引越ししたいけど、ラズパイのうちはラズパイにしかできないこともあるなあ。ラズパイだとおでかけもできるのか
- C'est le 3è jour de #VivaTech 2026, et le 3è épisode de la Quotidienne by @LesVoixdelaTech avec @benjaminvincent vous at
- @jeremyphoward @Zai_org Is it outputting Chinese characters during long conversations? That was at least a problem for m
- RT @elonmusk: People giving OpenClaw root access to their entire life https://t.co/jqW0FwiW03
- DietrichGebert/ponytail pushed `openclaw-ponytail-skill-sync` - suggesting integration with the openclaw agent framework — DietrichGebert/ponytail pushed openclaw-ponytail-skill-sync - suggesting integration with the openclaw agent framework. The repo accumulated 800+ stars without visible release triggers in commit...
- RT @gladstein: Three months ago I wrote my first reflections on AI I had been using OpenClaw for a few weeks, and it bl
- RT @empiriolabsai: Always-on OpenClaw or Hermes agents on EmpirioLabs. Simple monthly pricing, billed to credits, cancel
- And that's the most common thing in innovation. Small players have larger risk potential and hence they adopt new techno
- @leumas4samuel @jayhemz “Do you know how outlier works?” When I’m literally on OpenClaw Atlas after getting done with Ae
- RT @BlockView0214: 除了OpenClaw/Hermes，GitHub 上这波开源个人 agent，已经比想象中成熟： CowAgent（45k+ stars）：开源超级 AI 助手，能规划任务、跑工具、维护记忆和知识，前
- This community is for: - programmers - Indiehackers - Founder - AI / openclaw builders - iOS app builders — This community is for: - programmers - Indiehackers - Founder - AI / openclaw builders - iOS app builders
- Quick recap from ClawCon Nairobi yesterday, the first ClawCon on the African continent. 910 people registered. 400 show
- RT @murasametech: Raspberry Pi でOpenClawを運用する方法を以下の書籍で解説しています👇 https://t.co/pU1J65pa3o
- RT @wanerfu: 🔥 一条 AI 短剧，狂揽 41.8 万赞！ AI 短剧赛道，已经进化到“工业流水线”时代 爆款工作流：OpenClaw + XCrawl + Seedance 2.0 把短剧制作门槛，直接打到了地板上 详 — RT @wanerfu: 🔥 一条 AI 短剧，狂揽 41.8 万赞！ AI 短剧赛道，已经进化到“工业流水线”时代 爆款工作流：OpenClaw + XCrawl + Seedance 2.0 把短剧制作门槛，直接打到了地板上 详细 Skill 配置 + 操作步骤，…
- À VivaTech, Thibault Sottiaux (OpenAI) et Peter Steinberger (OpenClaw) racontent le tournant de l’IA agentique https://t
Releases
- openclaw 2026.6.9-beta.1 — 2026.6.9 Highlights - Richer Telegram delivery: Telegram now sends rich HTML, preserves rich markdown and sticker paths, renders progress drafts and command output more faithfully, and keeps mentions...
- OpenClaw repo stats for 2026-06-19 — 379454 stars, 79431 forks, 2614 contributors
- openclaw-cli 2026.6.8 on Homebrew — Your own personal AI assistant
Community
- fix(usage): exclude untimestamped records from daily ranges — Summary What problem does this PR solve? - Fixes ranged dashboard/session usage summaries so untimestamped cached transcript rows do not inflate selected-day totals or model usage. - Keeps all-time...
- fix(tailscale): reset stale serve entries before adding new ones on startup — Fixes #93936 Summary tailscale serve is append-only and never cleans up old entries. Each gateway restart with gateway.tailscale.mode: "serve" adds a new serve entry without removing stale ones,...
- Fix Discord default account startup priority — Fixes #77429 Real behavior proof - Behavior or issue addressed: Discord multi-account startup delay ordering previously followed the configured account id list directly, so a configured primary...
- fix(heartbeat): suppress private text delivery when heartbeat_respond tool is available but unused — Summary When heartbeat runs in message_tool_only mode and the model is given the heartbeat_respond tool but does NOT call it, the raw text response was being delivered to the Telegram channel —...
- docs(install): add vendor-neutral NVIDIA Jetson install guide — Summary Add a vendor-neutral NVIDIA Jetson install guide (docs/install/jetson.md) parallel to the existing raspberry-pi.md. Jetson modules (Orin, Xavier) are increasingly common always-on ARM64 edge...
- [Bug]: Skill loader ignores configured workspace path and hardcodes fallback to ~/.openclaw/workspace — Bug type Behavior bug (incorrect output/state without crash) Beta release blocker No Summary When a custom workspace directory is configured (e.g., on drive E:\), OpenClaw still attempts to load...
- [Bug]: Codex app-server session wedges indefinitely after a 429 rate-limit burst — terminal-idle watchdog still gated by `activeAppServerTurnRequests > 0` (regression/incomplete fix of #82681) — Bug type Behavior bug (incorrect state without crash) — session wedge / message loss. Summary On 2026.5.28, a Codex app-server session can wedge indefinitely with the agent silently not replying —...
- docs: vendor-neutral "Run OpenClaw on NVIDIA Jetson / edge hardware" guide — and how to reference turnkey devices? — Summary docs/install/ covers running the Gateway on Raspberry Pi and on specific VPS/cloud vendors (Fly, Hetzner, GCP, Azure, exe.dev…), but there's no guide for NVIDIA Jetson / dedicated edge...
- fix(auto-reply): guard context report tool entries — Summary - Guard cached context report tool entries before /context detail and /context map read tool names and size fields. - Reuse one normalized tool-report view for text diagnostics and the...
- fix(diagnostic): lower stall abort threshold and extend model-call recovery to CLI sessions — Summary The diagnostic heartbeat detects stalled sessions (model calls stuck >120s) but the abort threshold was 360s (6 minutes) — far too long to recover from a hung provider. Additionally,...
- fix(agents): guard message provider tool name reads — Summary - Guard filterToolsByMessageProvider() against tool descriptors whose name getter throws while provider allow/deny policy is applied. - Avoid reading descriptor names when no message-provider...
- fix(agents): guard lean tool name reads — Summary - Catch-bound local-model lean tool name normalization so unreadable descriptors cannot crash lean filtering. - Drop malformed or empty-name descriptors only while lean filtering is active;...
- fix(agents): guard deferred follow-up tool descriptors — Summary - Guard deferred follow-up guidance setup against active tools whose name getter throws. - Preserve unreadable-name tools in the returned list while still updating healthy exec/process...
- [codex] feat(mcp): stage binary tool content for relay — Summary Part of #87023. This adds a conservative relay foundation for MCP binary content: - stage MCP image, audio, and blob-backed resource content into OpenClaw-owned outbound media attachments -...
- fix(agents): guard code mode catalog tool names — Summary - Guard Code Mode catalog filtering against tools whose name getter throws before the shared Tool Search catalog compaction path runs. - Keep marked Code Mode control tools visible, drop...
- test(cron): cover cron base-session preservation during reaper sweep — Summary Adds regression coverage for the current cron session reaper boundary: expired canonical agent::cron::run: run-session descendants are pruned, while stable agent::cron: base rows are...
- fix(plugin-sdk): guard plain text tool name reads — Summary - Guard plain-text tool-call compatibility against unreadable tool name accessors in provider stream context. - Skip only malformed/unreadable context tool entries while preserving healthy...
- fix(cron): prune isolated-target cron sessions in session reaper — Summary - isCronRunSessionKey only matches cron session keys containing :run: (cron:jobId:run:runId). Isolated-target cron jobs (sessionTarget: isolated) generate base session keys without :run:...
- fix(text): strip commentary to=functions.* trace lines from assistant output (fixes #89668) — What does this PR do? Strips Anthropic-style commentary to=functions.* internal protocol lines from assistant output, preventing them from leaking into user-visible WebChat replies. Related Issue...
- ACP subagent deliver fails: streamTo silent stall + subagent_announce "No scope" fallback errors in 5.28 — Summary Two failure modes in ACP subagent announce/fallback path cause 100% delivery failure when the main session is fast (5–15s/turn). Both stem from the same root: the announce path does not...
- fix(agents): avoid enumerating wrapped tool descriptors — Summary - Stop before-tool-call and abort wrappers from spreading arbitrary active tool descriptors. - Add a shared wrapper descriptor copier that preserves the AgentTool contract fields plus...
- fix(providers): strip cache-boundary marker from non-Anthropic prompts — Summary - The internal SYSTEM_PROMPT_CACHE_BOUNDARY marker (\n\n) is meant to stay private to OpenClaw's Anthropic prompt-cache split. It currently leaks verbatim into the system prompt sent to...
- fix(control-ui): clear stale composer replays after send — Summary - After sending a message, the Control UI composer clears the draft. However, a stale handleInput event can replay the just-submitted text back into the textarea, making it appear as if the...
- [Bug]: Dashboard "今日模型调用" shows historical cumulative data instead of daily usage after v2026.5.28 upgrade — Bug type Behavior bug (incorrect output/state without crash) Beta release blocker No Summary After upgrading from v2026.5.27 to v2026.5.28, the Dashboard "Usage" page ("今日模型调用" / today's model calls)...
- fix(gateway): allow macOS app platform version refresh without re-pairing — Summary - macOS app enters infinite "Approve this Mac app to control OpenClaw" dialog loop after an OS update (e.g. macOS 26.5.0 → 26.5.1) - Root cause: resolvePinnedClientMetadata treats any macOS...
- Remove legacy tools.exec.security config in favor of tools.exec.mode — Summary tools.exec.security should be removed from the steady-state config surface and normalized into tools.exec.mode. Existing configs that still use the legacy key should be migrated by openclaw...
- fix(agents): snapshot client tool descriptors — Summary - Snapshot hosted/client tool descriptors once per attempt before reserved-name checks, Tool Search planning, allowlist collection, and registration. - Skip unreadable or malformed client...
- OpenClaw sponsors (107 total) — 107 sponsors supporting OpenClaw
- fix: show taxonomy on plugin cards (#2748) — Restore resolved plugin taxonomy on sorted catalog cards and show controlled category labels when author topics are absent. Prepared head SHA: eed82783f0ce23cc6744c0cde1bf2dd24dd1a804 Reviewed-by:...
- feat: surface plugin manifest icons (#2747)
- feat: add skill install backfill migration — Adds the aggregate skill install backfill migration, tests, and operator proof.
- feat: derive skill icons from categories (#2745)
- fix: persist registry backup seed completion state (#2746)
- ci: shard local-auth playwright gate (#2743)
- feat: enforce Convex retention policy — Add typed Convex retention policies, auth session/refresh-token cleanup crons, and retention guidance.
- feat: parallelize registry artifact backup backfill (#2741)
- feat: move topic discovery into search and category browse (#2732) — feat: search categories and topics fix: remove topic filters from browse sidebars feat: show top topics for selected categories fix(deps): resolve static audit advisories fix(search): scope skill...
- fix: stop auto-greeting staff emails (#2738)
- Add package runtime id repair command (#2734) — feat: add package runtime id repair chore: refresh audit gate overrides
- fix: shard skill stat event pruning
News
- sulpikar2/Qwen3.5-4B-Claude-4.6-Opus-Reasoning-openclaw
- Iter435/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...
- Tell HN: Anthropic no longer allowing Claude Code subscriptions to use OpenClaw
- Anthropic says OpenClaw-style Claude CLI usage is allowed again
- OpenClaw isn't fooling me. I remember MS-DOS
- AI Assistant Architecture: LLM, Memory, Tools, Routing, Observability — A production AI assistant is not "an LLM with a prompt". It is a system that accepts intent, keeps...
- How I connected OpenClaw to a Chinese frontier model and cut agent costs by 90% — I've been running OpenClaw for a few months. Love the agent workflow. Not so much the monthly...
- I think I found the first real reason to build AI agent workflows in OpenClaw — A practical case for using OpenClaw to automate receipt-to-ledger workflows: bounded inputs and outputs, clear human review gates, and a better split
- The journey to a healthcare AI personal assistant — Being unexpectedly immersed in the challenge of supporting a family member with a rare neurodegenerative condition is deeply distressing…Continue reading on Medium »
- Your AI Coding Assistant Should Still Be Working While You Sleep — The laptop was warm when I woke up.Continue reading on Medium »
- Building a Monthly Personal Review Agent with OpenClaw — Why I Built ItContinue reading on Medium »
- How to Use OpenClaw to Boost Your WordPress Workflow — Managing a WordPress website involves much more than publishing articles. You may need to research topics, write content, optimize for…Continue reading on Medium »
- Looking for an OpenClaw Alternative? Here Are 7 Worth Exploring in 2026 — AI agents have evolved far beyond simple chatbots. Tools like OpenClaw can automate tasks, remember context, interact with apps, and…Continue reading on Medium »
- Private GPT — Ask questions to your documents without an internet connection, using the power of LLMs. 100% private, no data leaves your execution environment at any point. You can ingest documents and ask...
- ChatSonic — Combining Google's knowledge graph, ChatSonic delivers real-time data, voice interaction, AI-produced visual content, personalized avatars, and API for services. It revisits past dialogs for informed...
- Reins — Empowering LLM researchers and hobbyists with seamless control over self-hosted models. Connect remotely, customize prompts, manage chats, and fine-tune configurations. All in one intuitive app.
- LlamaGPT — LlamaGPT is a chatbot that provides a ChatGPT-like experience, with no data leaving your device.
Don't miss what's next. Subscribe to Openclaw Newsletter: