ZeroClaw

ZeroClaw is a self-hosted personal AI assistant runtime — a single Rust binary you configure and run entirely on your own machine. It connects to 20+ LLM providers (Anthropic, OpenAI, Ollama, and more), reaches the world through 30+ channels (Discord, Telegram, Matrix, email, voice, webhooks, CLI), and acts through tools including shell, browser, HTTP, hardware peripherals, and custom MCP servers. Everything runs locally with your API keys, in your workspace — you own the agent, the data, and the machine.

• Multi-channel support — configure one agent to answer across Discord, Telegram, Matrix, email, webhooks, and CLI simultaneously, all delivered to the same agent loop.
• Provider-agnostic LLM routing — plug in Anthropic, OpenAI, local Ollama, or any OpenAI-compatible endpoint; fallback chains keep the agent running when a provider fails.
• Security-first with sandboxing — default supervised autonomy mode requires approval for medium-risk operations; OS-level sandboxes (Landlock, Bubblewrap, Seatbelt, Docker) and cryptographic tool receipts on every action.
• Hardware-capable — GPIO, I2C, SPI, and USB support for Raspberry Pi, STM32, Arduino, and ESP32 via the Peripheral trait.
• SOP engine — event-triggered Standard Operating Procedures via MQTT, webhook, cron, or peripheral events with approval gates and resumable runs.
• Web gateway and dashboard — HTTP/WebSocket gateway for clients with a built-in web dashboard for chat, memory browsing, config editing, cron management, and tool inspection.
• ACP (Agent Client Protocol) — IDE and editor integration via JSON-RPC 2.0 over stdio for seamless developer workflows.
• Minimal footprint — install via a single curl command or clone and build from source; minimal kernel-only binary is ~6.6 MB.
• TOML configuration — one config file at ~/.zeroclaw/config.toml covers providers, channels, security policy, and all runtime options.
• YOLO mode — opt-in fully autonomous mode for trusted dev environments where approval gates are not needed.