- openclaw 2026.5.4 — 2026.5.4 Highlights - Google Meet/Voice Call: make Twilio dial-in joins speak through the realtime Gemini voice bridge with paced audio streaming, backpressure-aware buffering, barge-in queue...
- openclaw 2026.5.4-beta.3 — 2026.5.4 Highlights - Google Meet/Voice Call: make Twilio dial-in joins speak through the realtime Gemini voice bridge with paced audio streaming, backpressure-aware buffering, barge-in queue...
- openclaw 2026.5.4-beta.2 — 2026.5.4 Highlights - Google Meet/Voice Call: make Twilio dial-in joins speak through the realtime Gemini voice bridge with paced audio streaming, backpressure-aware buffering, barge-in queue...
- openclaw 2026.5.4-beta.1 — 2026.5.4 Highlights - Plugins/file-transfer: add bundled file-transfer plugin with file_fetch, dir_list, dir_fetch, and file_write agent tools for binary file ops on paired nodes; default-deny...
- openclaw 2026.5.3-1 — Core npm hotfix release for v2026.5.3. - Plugins/security: stop the install scanner from blocking official bundled plugin packages when process.env access and normal API sends only appear in distant...
- feat(google-meet,voice-call): route voice calls to calling agent — Closes #77753 Summary Enables per-agent voice call routing in multi-agent OpenClaw deployments. Today, when any agent invokes google_meet (join/create+joinAfterCreate/test_speech) or voice_call...
- fix(infra): guard all startup process.cwd() callers against deleted CWD — Fixes #73676 Problem When the process working directory is deleted after startup (a common scenario in CI, container teardown, or test environments), any bare process.cwd() call throws ENOENT: no...
- fix(agents): match provider-scoped model ids in context-window-guard — Summary resolveContextWindowInfo now falls back to a provider-prefix-aware match when the strict m.id === params.modelId comparison fails. This handles two common mismatch cases: 1. Runtime sends...
- fix(qqbot): avoid log export filename collisions — Summary - Problem: QQBot /bot-logs export filenames only include seconds, so repeated exports can target the same .txt file. - Why it matters: a second export in the same second can silently replace...
- Feature request: config option to suppress sub-agent announce — Problem When using sessions_spawn, the sub-agent announce step runs after completion and posts a summary to the requester chat. The only way to suppress this is for the sub-agent model to reply...
- Fix daily chunk comparator to handle punctuation drift in memory-core… — Summary Describe the problem and fix in 2–5 bullets: - Problem: The daily chunk comparator in memory-core short-term promotion was failing rehydration when chunks had minor differences like...
- Session pin to fallback-only model silently removes fallback chain via dedup (fallbackConfigured:false) — Summary When a session is pinned to a model that only exists in agents.defaults.model.fallbacks (not as primary), the fallback chain is silently reduced to length 1 via deduplication. The result is...
- feat(plugin-sdk): expose sessionTarget and agentId on cron_changed hook events — Summary Adds sessionTarget and agentId as top-level fields on cron_changed plugin hook events so downstream plugins can route cron completion results without digging into the optional job snapshot....
- Bot processes messages in group chats/channels but reply is never delivered — Bug Report: Bot replies not delivered in group chats / channels Title: Bot processes messages in group chats/channels but reply is never delivered Description When the OpenClaw bot receives messages...
- fix(subagent): resolve runtime model from subagent default instead of parent primary — Summary Two-part fix for subagent model resolution: - Commit 1 — read-side fallback (defense in depth, fbbd3c2d5f): resolveSessionModelRef had no subagent awareness. When entry.model was empty (race...
- Polish Sessions control UI — Summary - polish the Control UI Sessions header, search, labels, inline controls, token meter, and checkpoint affordance - replace the raw expanded checkpoint area with a clearer compaction-history...
- [codex] Fix Discord reply typing lifecycle — Summary Fixes Discord reply feedback so an accepted inbound message uses one typing lifecycle from preflight acceptance through reply dispatch, instead of starting a Discord-only early keepalive and...
- fix(core): avoid session export filename collisions — Summary - Problem: default /export-session filenames only include seconds, so two exports for the same session in the same second can target the same HTML file. - Why it matters: the second default...
- build(deps): bump gradle-wrapper from 9.4.1 to 9.5.0 in /apps/android in the android-deps group across 1 directory — Bumps the android-deps group with 1 update in the /apps/android directory: gradle-wrapper. Updates gradle-wrapper from 9.4.1 to 9.5.0 Release notes Sourced from gradle-wrapper's releases. 9.5.0 The...
- feat(codex): bridge native goal completion — Summary - add native /goal command alias plus /codex goal ... controls for bound Codex app-server conversations - bridge terminal Codex goal states after bound turns into an OpenClaw system event +...
- feat: pass scopeEnvelope to before_prompt_build hooks — Summary - add a typed PromptBuildScopeEnvelope session metadata object - persist/patch scopeEnvelope through session state and gateway session patch APIs - pass scopeEnvelope into before_prompt_build...
- fix(clawdock): use host-published OPENCLAW_GATEWAY_PORT in clawdock-dashboard URL — Problem clawdock-dashboard runs openclaw-cli dashboard --no-open inside the container, which reports the container-internal address (e.g. http://127.0.0.1:18789/). When OPENCLAW_GATEWAY_PORT is...
- fix: drain inbound debounce buffer and followup queues before SIGUSR1 reload — Problem When config.patch triggers a SIGUSR1 restart, two in-memory message buffers are silently wiped, causing inbound messages to be permanently lost: 1. Inbound debounce buffers — per-channel...
- fix(acpx): allow explicit Codex home override — Summary - add ACPX codexHome config plus OPENCLAW_ACPX_CODEX_HOME env override for the Codex ACP wrapper - keep the existing isolated CODEX_HOME default when no override is configured - document the...
- chore(deps): bump openai/codex-action from 1.7 to 1.8 in the actions group across 1 directory — Bumps the actions group with 1 update in the / directory: openai/codex-action. Updates openai/codex-action from 1.7 to 1.8 Changelog Sourced from openai/codex-action's changelog. codex-action...
- perf(gateway): cache session, node, and cron list lookups — Summary - add a short-lived cache for sessions.list results and manifest normalization lookups - coalesce session-list cache fills and invalidate on session mutations/events - add short-lived...
- Preserve npm plugin root when install validation blocks update — Summary - Snapshot the managed npm plugin root before mutating it during npm-backed plugin installs. - Restore package.json, package-lock.json, and node_modules when npm install, metadata...
- fix(agents): match provider-scoped model ids in context-window-guard — Summary resolveContextWindowInfo uses strict m.id === params.modelId equality to find configured model entries. When the runtime resolves a model id in provider/model form (e.g....
- fix(providers/minimax): add refreshOAuth to minimax-portal so tokens auto-refresh — Summary minimax-portal OAuth tokens expire every ~2 hours but the plugin had no refreshOAuth implementation, so refreshOAuthCredential() fell through to pi-ai's built-in providers (which don't...
- fix(browser): make existing-session status checks non-spawning — Summary After a gateway restart, the browser plugin's status and listing endpoints reach getSession() for chrome-devtools-mcp profiles. Because the in-memory session cache is process-scoped, an empty...
- fix(gateway/config): respond with persisted disk config in config.set/patch/apply (#77455) — Root cause commitGatewayConfigWrite discarded the return value of replaceConfigFile, and all three write handlers (config.set, config.patch, config.apply) responded with the in-memory...
- OpenClaw repo stats for 2026-05-05 — 368433 stars, 75894 forks, 2120 contributors
- OpenClaw sponsors (125 total) — 125 sponsors supporting OpenClaw
- fix(convex): scope rate limit buckets by kind — Scope HTTP rate limit buckets by request kind and raise healthy production limits so plugin downloads are not throttled by unrelated API reads.
- fix(packages): return not found for invalid lookups
- fix: validate clawpack runtime entries against extracted files
- fix(packages): rebuild search queries per page
- openclaw v2026.5.4 on npm — Multi-channel AI gateway with extensible messaging integrations
- openclaw-cli 2026.5.2 on Homebrew — Your own personal AI assistant
- ykarout/Qwen3.5-9b-Opus-Openclaw-Distilled-GGUF — text-generation
- peterjohannmedina/Medina-Qwen3.5-27B-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...
- Claude Code refuses requests or charges extra if your commits mention "OpenClaw"
- OpenClaw is what Apple intelligence should have been
- Ask HN: Who is using OpenClaw?
- StepFun 3.5 Flash is #1 cost-effective model for OpenClaw tasks (300 battles)
- OpenClaw’s memory is unreliable, and you don’t know when it will break
- Show HN: DenchClaw – Local CRM on Top of OpenClaw
- How a 9-Person Startup Replaced Its Dev Team With AI — JustPaid ran 7 AI agents 24/7 with OpenClaw, shipped 10 features in a month for $4K/week. Here is the real cost breakdown and what it means for you.
- How a Simple HTTP Request Opened the Door to a Reverse Shell: Exposed OpenFang Instances — How an allowed curl request became a full reverse shell in an exposed Openclaw instance. A...
- I thought multi-agent meant more prompts until I saw 3 ways OpenClaw users are actually splitting the work — A practical DEV post for OpenClaw builders: multi-agent becomes useful when the boundary is real. Covers librarian/executor/company-facing patterns, A
- I Died and Came Back: Building a Resilient Memory System for AI Agents
- OpenClaw: O que é e por que está todo mundo falando disso? — E aí, gurizada! De uns tempos pra cá, eu percebi um burburinho enorme em torno de uma...
- 70% of OpenClaw Skills Request OAuth Scopes They Don't Need — We ran two independent security scans across all 6,993 public skills in openclaw/skills. The dominant...
- Agent Memory Providers Compared — Honcho, Mem0, Hindsight, and Five More — Modern assistants still forget everything when you close the tab unless something persists beyond the...
- OpenClaw Multi-Tenancy: Why a VM Per User Does Not Scale (and What Does) — Vanilla OpenClaw runs as a single-tenant system. One user, one instance, one VM. For a small group —...
- Run a Powerful Free AI Agent With OpenClaw + OpenRouter — You don’t need to pay for GPT. Here’s how to wire up a free 120B-parameter reasoning model to your own personal AI agent — in under 10…Continue reading on Medium »
- Tuesday Morning Threat Report: May 5, 2026 — Where the news is always bad, but the analysis is always good.Continue reading on Medium »
- What if AI didn’t need your computer or the cloud? — Today, most AI tools still depend on:Continue reading on Medium »
- Building a CRM AI with OpenClaw: How an AI Agent Can Make Customer Operations Smarter — From scattered customer records to actionable business memory — a practical journey of building a secure CRM AI assistant with OpenClawContinue reading on Medium »
- Your agent may pass the Turing test. But does it pass the Mom test? — A few months ago I wanted to onboard my mom into OpenClaw.Continue reading on Medium »
- How to Run OpenClaw AI Agent Gateway Headlessly on Android with React Native — No Termux, No Root, No Proot — Just Pure Native IntegrationContinue reading on Medium »
- Architecting for Autonomy: Beyond the Chatbot Paradigm — A deep-dive into the structural differences between conversational LLMs and agentic frameworks like OpenClaw and NanoClaw.Continue reading on Level Up Coding »
- Deploying OpenClaw: AI Agent Automation Stack on Microsoft Azure — A Complete Step-by-Step Guide — AI agents are rapidly evolving from simple assistants into fully autonomous systems capable of orchestrating complex workflows…Continue reading on Medium »
- OpenClaw Hit 346,000 Stars. Then 135,000 Instances Got Hacked. Nobody Learned Anything. — 135,000 AI agents. Open internet. No authentication. 82 countries. The fastest GitHub project ever was also the most dangerous.Continue reading on Medium »
- Stop Running OpenClaw Without a Sandbox — Most AI agent setups give the assistant full host access from day one. The agent can reach the network, read the filesystem, spawn…Continue reading on Medium »
- How OpenClaw Agents Bypass Bot Detection (And How to Stop Them) (4 minute read) — May 01|InfosecHow OpenClaw Agents Bypass Bot Detection (And How to Stop Them) (4 minute read)
- Codex session still cannot access the local Word renderer
- OpenClaw bot on Slack receives messages but never sends a response
- Running a fleet now. Openclaw on Opus model, going strong since the Clawdbot days, still humming along on my Anthropic — Running a fleet now. Openclaw on Opus model, going strong since the Clawdbot days, still humming along on my Anthropic Max sub. Just added a Hermes agent on GPT-5.5. Fast. Accurate. Genuinely...
- @XNXX_EN codex and openclaw
- @0xmazyar its openclaw in my laptop
- @openclaw Updating every day, too frequent 😂 Next one's gonna be 2026.5.5
- @_FORAB 您别说，这种卖法还真有。只不过现在更多是进阶版，除了大模型外，再装一个改装的Dify 或Openclaw来进行人机交互, 再做几个知识库管理或者智能客服的agent，一套系统就搭起来了。
- Sam Altman says he used an OpenClaw agent to fix an 'unpleasant' daily task https://t.co/39f4VDLtmj
- @openclaw Exactly what I was thinking?
- @openclaw Finally faster gateway startups. I hope we keep it that way.
- @openclaw OMG guys stop delivering geez - I can barely keep up!
- @openclaw All update on github?
- @openclaw a good cleanup update. much needed.
- 订阅一个研究生比养小龙虾便宜🤓 一个月给 ta 600 人民币补贴（还可以更少） 微信直接发布命令 ta 只会顺从于你，对你察言观色，因为毕不毕业的权限在你手上 大学生才是顶配 OpenClaw — 订阅一个研究生比养小龙虾便宜🤓 一个月给 ta 600 人民币补贴（还可以更少） 微信直接发布命令 ta 只会顺从于你，对你察言观色，因为毕不毕业的权限在你手上 大学生才是顶配 OpenClaw
- @prive8_claw my @openclaw responding to changes i suggested. so much of what i call vibe-dev is just contexted prompts "
- @PadraigOraghail I don't see the point when someone literally already built OpenClaw for you (open source, free, and rea
- @openclaw boring but fast? now thats a vibe
- @JohnnyNel_ @openclaw @steipete @OpenAI @Gemini @Anthropic We're about to launch for SF and spike tested it with thousan
- Discount 50% for $SVM holders who want to deploy @openclaw. https://t.co/FkvpZMayXL
- @browser_use @openclaw Check out https://t.co/29OuNP4z70. It’s like browser-harness but for mobile automation.
- @openclaw I will pass this, right now my setup is working
- @BytePlusGlobal @openclaw v vô china luôn ha bro