This project lets you run an AI assistant on your own infrastructure, whether that's a laptop, a VPS, or a home server. It integrates with the messaging apps you already use — Discord, WhatsApp, Telegram, Slack, Signal, iMessage, and many others — and also supports voice interaction and hosted widgets. The Gateway component acts as a local control plane that handles session management, tool execution, event processing, and channel connections. You can connect it to hosted model providers like Anthropic or OpenAI, or run local models via Ollama, with per-agent routing and failover. All state, memory, and credentials remain on your hardware. The project is governed by the OpenClaw Foundation, an independent non-profit, and has no paid tiers, hosted services, or tokens. The codebase is MIT licensed.
Should you self-host openclaw?
openclaw scores 93/100 on selfhost.directory (how scoring works) and ranks #1 of 123 in AI. It has 394 532 GitHub stars, +2 828 in the last 30 days.
It shipped 44 stable releases in the past year. The latest, v2026.11.3, came out on 10 November 2026. New versions land roughly every 2 days.
Setup is moderate: expect some configuration beyond a single command. An official Docker image is available. It is MIT-licensed and written mostly in Typescript.
OpenClaw 🦞 — Your assistant, on your devices, in your chats
OpenClaw is an open-source AI assistant that runs on your own computer and meets you in the channels you already use: Discord, iMessage, Slack, Teams, Telegram, WhatsApp, and 20+ more, plus native apps for macOS, iOS, Android, Windows, and Linux. One Gateway runs it as a personal claw on a laptop or as a shared multiplayer claw; configuration is the only difference.
Yours, with no catch. State, memory, and credentials live on your hardware. Models and agent harnesses (Claude, Codex, local models) are plugins you can swap without changing anything else. Your prompts go to the model provider and chat platforms you configure, plus any diagnostics export you enable yourself; by default OpenClaw itself phones home for nothing but a daily version check, anonymous feature statistics are opt-in, and update.checkOnStart: false disables both (what OpenClaw sends). OpenClaw is stewarded by the OpenClaw Foundation, an independent 501(c)(3), and has no paid tier, hosted service, or token. The architecture case — trusted gateway, untrusted execution, policy enforced by code — is in Why OpenClaw.
That command is for npm 12 or npm 11.16+. On npm 11.15 and earlier, omit
--allow-scripts=openclaw. See the
installation guide for the lifecycle script
contract, Docker, Nix, and other deployment paths.
Quick start
On a fresh install, the installer scripts start onboarding automatically.
Complete the wizard they open. If you installed the package directly with npm,
pnpm, or Bun, run:
openclaw onboard --install-daemon
After onboarding:
openclaw gateway status
openclaw dashboard
Onboarding verifies model access, creates the workspace, and configures the Gateway. The last command opens the Control UI; send a message there to confirm the assistant is working. See the getting started guide for channel setup and troubleshooting.
How it fits together
The Gateway is the local control plane for sessions, tools, events, and channel connections.
The Control UI, CLI, and TUI connect to the Gateway.
Channels bring the assistant to WhatsApp, Telegram, Slack, Discord, Google Chat, Signal, iMessage, and other messaging services.
Companion apps and nodes add voice, Canvas, camera, screen, and device-local actions on supported platforms.
Treat inbound messages as untrusted input. DM-capable channels pair unknown senders by default; approve a pairing request with openclaw pairing approve <channel> <code>.
Tools run on the host for the main session unless you configure sandboxing. Read the security guide, exposure runbook, and sandboxing guide before connecting other users or exposing the Gateway remotely.
OpenClaw is developed in the open by the OpenClaw Foundation, an independent 501(c)(3). The Foundation employs the core team and signs releases. Donors and infrastructure sponsors support the Foundation; none of them own or direct the project. OpenAI is a donor, not an owner.
Community
See CONTRIBUTING.md for maintainers and contribution guidelines; AI-assisted PRs are welcome.
Use the issue chooser for bugs and feature requests, get official OpenClaw support at community.openclaw.ai, and report vulnerabilities through SECURITY.md. New capabilities usually belong in plugins built on the plugin SDK and shared through ClawHub.
Special thanks to Mario Zechner for his support and for pi, and to Adam Doppelt for the lobster.bot domain.
Donors and sponsors
The Foundation is funded by donors including Amazon, Lobster Computer Company, Offline Holdings, OpenAI, Red Hat, and the University of Michigan, with infrastructure support from Blacksmith, Convex, GitHub, NVIDIA, and Vercel.
AI summary Private task results and improved task handoffs
2026.11.3
Highlights
Private task results: Keep private subagent completions out of streamed chat replies. (#188490) Thanks @fuller-stack-dev.
Reliable child-task handoffs: Allow a parent to hand off to replacement work after cancelling a child, without leaving the cancelled child's completion receipt blocking the handoff. (#178893; related #178565) Thanks @brianmcguire and @letsc4.
Visible tool failures: Report unanswered tool failures instead of silently ending a user-addressed turn as a success. (#182458; related #99645) Authored by @steipete. Thanks @rdleclerc.
Bounded authentication: Stop stalled provider authentication at the run deadline and propagate cancellation through authentication work. (#189144) Thanks @amittell.
Sandboxed image access: Keep required-sandbox image previews inside allowed folders, even in Full Access mode, while preserving explicit administrator image allowances. (#176203) Thanks @shakkernerd.
Verified workspace connections: Preserve saved node certificate pins when an empty fingerprint is supplied, and verify workspace-transfer certificates before sending credentials or data. (#125646, #190002) Thanks @wanyongstar and @avelev99.
Faster, responsive Gateway: Reduce startup module-file loading overhead and avoid stalls when abort-command detection encounters long punctuation runs. (#192217, #184433) Thanks @karkarl, @RomneyDa, @LeizhenZhang, and @obviyus.
Changes
No new configuration options or features.
Fixes
Credential isolation: Keep reused and cached plugin results scoped to their connection credentials, and keep other plugins' transport hooks on the provider's origin. (#178444, #177974, #189423) Authored by @steipete. Thanks @obviyus.
Search and memory credentials: Keep Firecrawl search and scrape keys with their scoped endpoints and embedding credentials with their destination. (#190706, #184966) Thanks @zhangguiping-xydt and @eleqtrizit.
Credential redaction: Redact inline environment credentials from configuration reads and reflected search credentials from Exa errors. (#190618, #178745) Authored by @steipete. Thanks @yetval.
Child-process environments: Prevent managed Signal daemons and bundled language servers from inheriting unrelated Gateway secrets. (#179198, #179415) Thanks @VACInc.
Authentication errors: Surface credential-read failures rather than treating them as missing credentials, and reject explicitly selected auth profiles belonging to incompatible providers. (#173970, #181597) Authored by @steipete. Thanks @vincentkoc.
Channel safety: Keep incoming Twitch IRC content out of parser error logs and require explicit request IDs when multiple Tlon approvals are pending. (#174552, #186243) Authored by @steipete. Thanks @mmaps.
Scheduled jobs: Handle queued cron storage failures without exiting the Gateway, and clear one-shot cleanup when a job becomes recurring. (#175678, #168985) Authored by @steipete. Thanks @yxshee, @wangmiao0668000666, and @obviyus.
Updates: Let profile-backed updates recognize their own maintenance lock and preserve cancellation when ledger writes are deferred. (#177792, #187998) Thanks @GrahamMcBain and @vincentkoc.
Update recovery: Recover safely when the initial package-presence check times out, requiring a fresh valid directory identity before replacement and retaining rollback to the original package. (#181906) Authored by @steipete.
Anthropic usage caps: Recognize exhausted plural usage-limit errors without spending the transient rate-limit retry budget. (#176186; related #176179) Thanks @sunlit-deng and @blue-dot-1.
Terminal sessions: Forward /reset soft to the Gateway instead of hard-resetting the session. (#172431) Thanks @yetval and @obviyus.
Git recovery: Preserve uncertain command failures during Git repair rather than treating an unconfirmed operation as successful. (#176006) Authored by @steipete.
This audited record covers the complete v2026.11.2..abf11ef history: 34 in-range PRs + 0 retained seed-only PRs = 34 unique PRs. The generation manifest also supplies direct commits as editorial input; the grouped notes above prioritize user impact.
This is a gateway-only extended-stable release, published from a retired maintenance line by explicit maintainer decision. This release is OpenClaw from the end of September 2026, plus critical security updates, reliability and performance fixes, and features like new model support. The latest version of OpenClaw at the time of this release is 2026.11.2
2026.9.37
Highlights
Private inspection stays private: paired-node photos and screenshots are no longer automatically attached to replies, and explicit media-read denials are honored when preparing reply attachments. (#174814, #184964) Thanks @eleqtrizit.
Linked accounts survive resets: sessions keep their person-linked model account through /new, /reset, and idle or daily rollover. Fixes #179432. (#179433) Thanks @masatohoshino Authored by @steipete. Thanks @masatohoshino.
Connections recover cleanly: channel accounts can recover after a manual stop times out, and ClickClack reconnects after silent network drops. Fixes #184852. (#177444, #184853) Thanks @masatohoshino, @YangManBOBO, and @obviyus.
Windows state access works across path aliases: equivalent Windows paths preserve the same state ownership instead of falsely reporting offline maintenance. Fixes #176163. (#176174) Thanks @tjarrau-dev Authored by @steipete. Thanks @tjarrau-dev.
Changes
Faster plugin admission: reuse native-reference validation within the same admission capture instead of repeatedly walking companion directories, while revalidating replacement captures and recovery copies. Fixes #162047. (#162254) Thanks @WG-Mojo Authored by @steipete. Thanks @WG-Mojo.
Fixes
Private diagnostics: omit suppressed reply text and background command contents from verbose diagnostics while retaining status metadata. Thanks @steipete.
Media boundaries: keep paired-node inspection photos and screenshots private by default, and honor explicit read denials for reply attachments without blocking authorized managed outputs. (#174814, #184964) Thanks @eleqtrizit.
Discovery output: escape terminal control characters in untrusted Gateway discovery fields. (#183355) Thanks @Alix-007.
State and write outcomes: recognize committed SQLite writes whose worker result cannot be serialized, avoiding unsafe replay, and preserve Windows ownership across equivalent path spellings. Fixes #176163. (#177321, #176174) Thanks @tjarrau-dev Authored by @steipete. Thanks @tjarrau-dev.
Session accounts: retain linked model-account selections and their provenance through explicit resets and automatic rollover. Fixes #179432. (#179433) Thanks @masatohoshino Authored by @steipete. Thanks @masatohoshino.
Channel recovery: recover channel accounts after timed-out manual stops and restore ClickClack message reception after silent network drops. Fixes #184852. (#177444, #184853) Thanks @masatohoshino, @YangManBOBO, and @obviyus.
Local tool connections: replace unexpectedly closed MCP loopback listeners and retain their cleanup work for shutdown rather than leaving later tool calls permanently disconnected. Fixes #139340. (#183352) Thanks @SC-10749 and @eli-pavlov Authored by @steipete. Thanks @eli-pavlov and @SC-10749.
Command results and temporary files: preserve a completed command's exit result when detached-descendant cleanup is uncertain, retain a visible cleanup warning, and remove temporary MCP configuration directories when writing fails. Fixes #179555. (#185636, #179603) Thanks @xiaopan-xydt and @Yolo521-bot Authored by @steipete. Thanks @Yolo521-bot and @xiaopan-xydt.
Gateway credentials: keep local password SecretRefs active when a separate remote token is configured, and retain a saved node TLS pin when the fingerprint option is empty or whitespace-only. (#128175, #125646) Thanks @TheAngryPit, @CassieMei, @obviyus, @RomneyDa, and @wanyongstar.
Update and restart reliability: preserve rollback snapshots after interrupted writes, release completed maintenance before reopening requester policy, confirm predecessor stop receipts during Doctor maintenance, and prepare configuration restarts outside the restart-signal fence. (#189946, #177792, #187104, #172777) Thanks @kaguranananana, @GrahamMcBain, and @RomneyDa Authored by @steipete. Thanks @RomneyDa, @kaguranananana, and @GrahamMcBain.
Release artifact verification: qualify the candidate's exact plugin artifacts, route validation through the admitted verifier, and accept successful admission jobs that omit step details while preserving run, attempt, artifact, and authority checks. (#164464, #173437, #188597) Thanks @RomneyDa.
Complete contribution record
This audited record covers the complete v2026.9.36..2f7023b history: 28 in-range PRs + 0 retained seed-only PRs = 28 unique PRs. The generation manifest also supplies direct commits as editorial input; the grouped notes above prioritize user impact.
Android APK: skipped — apps/android/version.json is 2026.8.2, release train is 2026.11.2; run pnpm android:version:pin -- --from-gateway before the next tag.
Android End Talk with a compatible Gateway stops listening without canceling accepted agent work, so its answer can still arrive in chat; explicit cancellation remains available. (#172276, #188798) Thanks @RomneyDa.
Provider recovery respects provider-specified rate-limit timing, and Gemini Flash text models use Interactions by default on the official API-key endpoint while custom proxies retain their existing default. (#180280, #189395) Thanks Ayaan Zaidi, @evomach, @sukhdeepjohar, @RomneyDa.
Changes
Claude CLI: saved subscription accounts can make configured models available without a separate native login, and operators can opt into native Agent and Workflow orchestration through plugins.entries.anthropic.config.managedNativeTools; the default remains off and restricted runs keep their existing tool limits. (#179203, #183765) Thanks @RomneyDa.
Anthropic: API-key setup can record an optional Anthropic workspace ID; existing workspace headers are preserved, and workspace-scoped keys need no extra setting. (#186976) Thanks Ayaan Zaidi, @evomach, @sukhdeepjohar, @RomneyDa.
Google: Interactions requests honor managed proxy and TLS settings, rotate eligible API keys before streaming after a rate limit, preserve native video input, and forward nucleus sampling consistently; choose google-generative-ai when explicit Gemini cache handles are required. (#189395, #190695) Thanks @RomneyDa.
Control UI: the model picker shows the configured active context budget rather than only the model maximum, and short session links retain their intended navigation target. (#180870, #183148) Thanks @RomneyDa.
Google Chat: remote HTTP(S) attachments are delivered as visible links alongside captions; local files and native uploads remain unsupported with service-account authentication. Thanks @RomneyDa.
Update diagnostics: detailed readiness separates a ready core from channel outages, allowing repaired updaters to finish with named channel warnings without hiding HTTP 503 or weakening normal service-start checks. Thanks @duyufansh2008, @RomneyDa.
Fixes
Network and tool safety: ClickClack and ClawHub reject redirects into unrelated private services, Discord CDN downloads no longer inherit browser private-network access, and FIFO native-hook credentials are rejected without hanging the agent. (#190944, #190939, #191458, #191423) Thanks @pgondhi987, @SebTardif, and @Leon-SK668.
Log redaction: malformed custom entries no longer disable valid rules or built-in secret protection; invalid policy is reported while safe redaction remains active. (#173471) Thanks @skyzhao1223 and @RomneyDa.
Secret egress: reject chains relying on the distrusted emSign ECC G3 root in the managed proxy and exported subprocess CA bundle; restart the Gateway after updating, and update clients with independent trust stores separately. (#179249) Thanks @RomneyDa.
macOS connectivity: configured credentials take precedence over stale app environment, typed environment secret references resolve correctly, and remote Settings respects the Control UI base path. (#191468, #185141, #128175) Fixes #164596 Thanks @RomneyDa, @anmol0b, and @goldmar.
Mac voice and embedded pages: avoid duplicate voice sends and false sent feedback, and show a recoverable load error instead of a blank or stale Mac tab. (#186271, #185432) Thanks @anmol0b, @etzelm, and @RomneyDa.
Native voice and capture: iOS and macOS Talk tolerate slow audio acknowledgments within bounded buffering, long meeting replies are not cut off merely because generation outruns playback, and oversized Mac recordings fail with size guidance instead of dropping the node connection. Thanks @MertBasar0, @anmol0b, @shakkernerd, @kybrcore, @sukhdeepjohar, @potoior, and @RomneyDa.
Android and desktop compatibility: Android Chat waits for the role-allowed agent list after switching Gateways, and independently updated Mac nodes can return optional cursor coordinates without a computer-use contract error. Thanks @MertBasar0, @anmol0b, @shakkernerd, @kybrcore, @sukhdeepjohar, @potoior, @RomneyDa.
Channel recovery: Slack detects dead relay connections without treating ordinary silence as failure, Tlon reconnects stalled streams so unacknowledged events can replay, and managed Tailscale exposure recovers after a daemon restart while local authenticated access stays available. Thanks @RomneyDa.
Channel delivery: Feishu keeps text from before and after tool calls in the same streaming card; Synology confirms delivery only from recipient-matched message IDs; Twitch targets and Zalo unsupported-event text follow their channel handling rather than silently losing usable content. Thanks @RomneyDa.
Channel credentials and notices: implicit Slack default accounts retain complete top-level credentials alongside named accounts, channel secret consumers respect account activation, and Nostr reports failed pairing notifications without undoing approval. Thanks @RomneyDa.
iMessage storage: recovery cursor keys no longer contain a PostgreSQL-incompatible NUL separator, allowing the channel to start with PostgreSQL-backed state. (#184919) Fixes #184862 Thanks @duyufansh2008, @RomneyDa, and @etzelm.
Session and agent authority: deleting one agent no longer breaks other agents' tool authority, and local session-capability changes stop inappropriate provider fallback instead of trying every model. (#174918, #179312) Fixes #179290 Thanks @steipete, @chelsealong, @obviyus, and @Marvinthebored.
Prompt caching: session tool definitions stay stable across turns, including completions, heartbeats, and non-owner requests, avoiding unnecessary prompt-cache misses. (#175309, #176167) Thanks @obviyus.
Conversation recovery: compaction retains the active human request across repeated summaries; queue overflow summaries retain each visible sender and conversation context; exec completion guidance survives overflow. (#180214, #187833, #190698, #182188, #184216) Thanks @RomneyDa and @aniketkrs.
Codex continuity: conversation-wide restrictions retain compatible native threads, temporary restrictions preserve the ordinary conversation binding, and a missing thread with recorded history reports an error instead of silently starting over. Thanks @skyzhao1223 and @RomneyDa.
Codex replies: transient retries can summarize the original request and completed tools without rerunning them; terminal handoffs allow already-submitted steering a bounded chance to confirm consumption and do not automatically replay uncertain input. (#183077, #184062, #181839, #167390) Thanks @RomneyDa.
Shared credentials and model availability: concurrent MCP callers can reuse a fresh OAuth token without unnecessary exclusive locking, rate-limit cooldowns follow provider timing instead of accumulating hours of local backoff, and fresh model publications replace superseded refresh results. (#179997, #180280, #179144) Thanks @RomneyDa.
Claude CLI and scheduled work: canceled cron runs stop reply hooks, interrupted long-running jobs no longer prevent scheduler startup, oversized or invalid CLI streams cancel their subprocess promptly, and managed completions retain the policy-filtered tool catalog. (#181245, #191422, #159931, #172026) Thanks @steipete and @fuller-stack-dev.
Session history and memory: long transcript decoding releases database read snapshots promptly, bounded history pages keep their requested anchors and adjacent messages, and memory indexing waits for archived transcript materialization. (#164442, #172970, #177937, #182091, #190182, #189284, #188450, #188483, #190690) Fixes #171467 Thanks @RomneyDa, and @etzelm.
Subagent and restart recovery: live descendants keep their expired parents available for communication, accepted yields settle sibling tool results before handoff, and stale interrupted input no longer blocks a new human message indefinitely. (#179057, #184021, #180853, #181871, #178888, #184041, #184069, #172959) Thanks @RomneyDa.
Agent handoffs: waiting session sends recover the stored long reply instead of returning only a short lifecycle preview, and ACP-created sessions retain the requested working directory for later prompts. (#178271, #176852) Thanks @RomneyDa.
Control UI recovery: silently dead Gateway connections reconnect without replaying uncertain sends, consumed slash-steering messages leave the queue, and routine restart notices stay concise. (#179101, #178642, #180024, #172843) Thanks @RomneyDa.
Plugin lifecycle: startup avoids recopying managed binaries, repairs owned stale host links after installation-root changes, and keeps plugin background readers alive for admitted callbacks and cleanup; internal plugin reload resumes stay in the current session rather than using an old external recipient. (#187865, #181140, #182207, #179031, #181270, #179549, #190982) Thanks @Satoshi-Kusumoto, @RomneyDa, @jasonftl, and @obviyus.
Update startup and migration: recover with retained worker indexes, avoid crashes in automatic migration recovery from 2026.10.7–2026.10.9, preserve recovery errors, and reject Git targets that require an unavailable newer Node before staging. (#189023, #189713, #191495, #181715) Thanks @RomneyDa and @steipete.
Configuration and store recovery: explain how to recover when an older binary meets newer config; relative session stores resolve against the active state directory, and Doctor merges legacy misplaced stores with backups while preserving conflicting conversations and refusing unsafe retirement. (#187639, #182444, #181554, #180414, #179089) Fixes #152791 Thanks @RomneyDa, @duyufansh2008, @steipete, and @kip-claw.
Service and backup reliability: repaired updaters distinguish durable database changes from read-only WAL bookkeeping, wait for the serving Mac child to release state, preserve valid replacement config without metadata, and keep unrelated service overrides intact during supported Node repairs. Thanks @RomneyDa.
This audited record covers the complete v2026.10.10..63e490a history: 78 in-range PRs + 0 retained seed-only PRs = 78 unique PRs. The generation manifest also supplies direct commits as editorial input; the grouped notes above prioritize user impact.
Android APK: skipped — apps/android/version.json is 2026.8.2, release train is 2026.11.1; run pnpm android:version:pin -- --from-gateway before the next tag.
Android APK: skipped — apps/android/version.json is 2026.8.2, release train is 2026.10.10; run pnpm android:version:pin -- --from-gateway before the next tag.
0 comments