Back to directory

ZeroClaw

A self-hosted personal AI agent runtime, distributed as a single Rust binary, that connects to about twenty LLM providers and reaches the user through more than 30 channels such as Discord, Telegram, email and a custom CLI.

ZeroClaw

ZeroClaw runs entirely on the user's own machine with their own API keys and workspace. It talks to model providers including Anthropic, OpenAI and Ollama, acts through tools such as a shell, browser, HTTP client, custom MCP servers, and hardware peripherals over GPIO, I2C, SPI or USB on boards like Raspberry Pi, STM32, Arduino and ESP32, and answers across whichever configured channel a message arrives on -- Discord, Telegram, Matrix, email, webhooks or its own CLI -- funneling everything into one agent loop.

Its default autonomy level is supervised: medium-risk operations require approval and high-risk ones are blocked, enforced through workspace boundaries, a command policy, OS-level sandboxes such as Landlock, Bubblewrap, Seatbelt or Docker, and a cryptographic receipt recorded for every action, with an opt-in unrestricted mode for trusted development machines. An HTTP/WebSocket gateway and web dashboard expose chat, memory browsing, configuration editing and cron management, an event-triggered Standard Operating Procedure engine runs approval-gated, resumable workflows, and an Agent Client Protocol integration connects it to IDEs and editors.

It is configured through a single TOML file, is written in Rust, and is licensed under Apache-2.0.

More in Agent frameworks

See all 902
  • Recurse

    recurse.run

    A serverless harness that lets a coding agent design, test, and deploy its own specialist sub-agents as tools, MCP servers, or bots, defined by a schema-validated manifest.

  • ProofForge

    shulgin.is-a.dev

    An AI-agent pipeline that decomposes a mathematics problem, proves each piece, and formalizes the result in Lean 4 and Mathlib until the Lean kernel accepts it, with six of its proofs already merged into Google DeepMind's formal-conjectures repository.

  • local-coder

    github.com

    Builds a local coding-agent team on top of Ollama and OpenCode by detecting the machine's hardware, recommending models for seven specialist roles, and testing whether each assigned model can actually use tools, edit a repository and pass verification.