OpenClaw is a personal AI assistant platform written in TypeScript that runs on user-owned devices across any operating system and platform. The project emphasizes data ownership and local-first operation, with a distinctive branding around the lobster theme. The assistant integrates with a comprehensive set of messaging channels including WhatsApp, Telegram, Slack, Discord, Google Chat, Signal, iMessage, IRC, Microsoft Teams, Matrix, Feishu, LINE, Mattermost, Nextcloud Talk, Nostr, Synology Chat, Tlon, Twitch, Zalo, WeChat, QQ, and WebChat, allowing users to interact with their AI assistant through platforms they already use daily.
The core architecture centers on a local-first Gateway that serves as a control plane managing sessions, channels, tools, and events. Users can set up OpenClaw through the recommended onboarding process via the openclaw onboard command, which guides configuration of the gateway, workspace, channels, and skills across macOS, Linux, and Windows. The project supports multiple installation methods including npm, pnpm, and bun, with Node 24 recommended or Node 22.19 and above as minimum requirements. A native Windows Hub companion app provides setup assistance, tray status monitoring, chat functionality, and local MCP mode for Windows desktop users.
OpenClaw includes sophisticated multi-agent routing capabilities that allow users to route inbound channels, accounts, and peers to isolated agents with separate workspaces and per-agent sessions. The platform features voice integration through Wake Mode and Talk Mode, supporting wake words on macOS and iOS with continuous voice capability on Android, backed by ElevenLabs and system TTS fallback. A Live Canvas provides an agent-driven visual workspace with A2UI support. First-class tools include browser access, canvas rendering, nodes, cron scheduling, session management, and Discord and Slack action integration.
The repository shows active development with 7086 open issues as of the latest tracking period, reflecting an engaged user base and ongoing feature development. GitGenius activity tracking reveals median issue and pull request response latency of 0.7 hours with a mean of 21.2 hours across 10000 items, indicating responsive maintainership. The most prevalent issue labels are bug with 4715 occurrences, stale with 2695, and enhancement with 1958, showing a focus on stability and feature expansion. Top contributors steipete, sebslight, and vincentkoc have driven significant development momentum with 25825, 8825, and 4543 tracked events respectively. The project shares contributor overlap with major repositories including microsoft/vscode, anthropics/claude-code, and microsoft/typescript, indicating cross-pollination with significant open-source ecosystems.
Security is a primary design consideration, with default behavior treating inbound direct messages as untrusted input. The platform implements DM pairing policies requiring approval codes for unknown senders on major platforms, with optional open policies for explicitly configured allowlists. Sandboxing support allows non-main sessions to run in isolated environments using Docker, SSH, or OpenShell backends, with configurable tool access restrictions. The project is backed by sponsors including OpenAI, GitHub, NVIDIA, Vercel, Blacksmith, and Convex, supporting its infrastructure and development goals.
