ZeroClaw is an open-source personal AI agent designed to run on hardware you control. Distributed as a lightweight Rust binary, it works with your API keys or local models, without requiring a cloud subscription or provider account. The project supports more than 70 language model providers, including OpenAI, Anthropic, Google Gemini, and local options such as Ollama, LM Studio, llama.cpp, and vLLM. It connects to over 30 channels, including Telegram, Discord, WhatsApp, Slack, Signal, email, and Matrix, letting users interact with the same agent across enabled services.

ZeroClaw offers guided setup, interactive chat, one-shot prompts, memory and tool support, and runtime services for gateways, daemons, webhooks, and scheduled workflows. It is built for systems ranging from workstations and cloud virtual machines to Raspberry Pi devices, with Docker support. Security features include supervised actions, approval requirements, command allowlists, workspace scoping, and operating-system sandboxing. Optional HMAC-SHA256 tool receipts provide verifiable records of successful calls. Released under MIT or Apache-2.0 licenses, ZeroClaw also supports configurable, pluggable components.