zeroclaw

ZeroClaw is a personal AI assistant infrastructure distributed as a single Rust binary. It functions as an agent runtime that you configure and run on your own hardware, retaining full ownership of the agent, the data, and the machine it operates on. Core capabilities include: - Multi-channel communication: One agent can answer across 30+ configured channels. Inbound messages from Discord, Telegram, Matrix, email, webhooks, and CLI are all delivered to the same agent loop. - Provider-agnostic LLM support: Pluggable model providers include Anthropic, OpenAI, local Ollama, and any OpenAI-compatible endpoint. Fallback chains and routing keep the agent running if a provider fails. - Security-first design: Default autonomy mode is supervised, where medium-risk operations require approval and high-risk operations are blocked. Features include workspace boundaries, command policy, OS-level sandboxes (Landlock, Bubblewrap, Seatbelt, Docker), and cryptographic tool receipts on every action. A YOLO mode exists for trusted development environments. - Hardware capability: GPIO, I2C, SPI, and USB support for Raspberry Pi, STM32, Arduino, and ESP32 via the Peripheral trait. - Gateway and dashboard: An HTTP/WebSocket gateway for clients with a web dashboard for chat, memory browsing, config editing, cron management, and tool inspection. - SOP engine: Event-triggered Standard Operating Procedures via MQTT, webhook, cron, or peripheral events, with approval gates and resumable runs. - IDE integration: Agent Client Protocol support via JSON-RPC 2.0 over stdio for editor integration. Configuration is managed through a single TOML file. The runtime uses SQLite and embeddings for memory. It can be installed as a system service via systemd, launchctl, or Windows Service. The project is dual-licensed under MIT and Apache 2.0.