thClaws/thClaws

Open-source AI agent harness in native Rust — GUI, CLI, headless, and webapp from one binary. Multi-provider, MCP, skills, plugins, agent teams.

What it solves

thClaws is an open-source AI agent workspace designed for local execution. It provides a unified environment where users can code, automate tasks, and manage knowledge, while maintaining sovereignty over their data and API keys. It solves the problem of fragmented AI tools by combining a terminal, a GUI, and agent orchestration into a single binary.

How it works

Built in Rust, the system uses a single engine that powers four different interfaces: a Desktop GUI, a CLI REPL, a non-interactive mode for scripts, and a Webapp. It integrates with a vast array of LLM providers (including local models via Ollama) and supports the Model Context Protocol (MCP) for extending tool capabilities. The agent loop handles turns, tool calls, and context management, while OS-level confinement (like macOS Seatbelt or Linux Landlock) ensures that shell commands are executed within a secure filesystem boundary.

Who it’s for

Developers, power users, and enterprises who want a highly customizable, local-first AI agent harness that can automate complex workflows, manage project-specific knowledge bases, and integrate with various LLM providers without being locked into a walled garden.

Highlights

  • Multi-Surface Interface: Access the agent via Desktop GUI, CLI, or a web-based API surface.
  • Extensive Provider Support: Compatible with 1,100+ models across 27 providers, including local-first options.
  • Secure Execution: OS-enforced filesystem boundaries for shell commands to prevent unauthorized access.
  • Agent Orchestration: Supports model-driven subagents, concurrent side-channel agents, and multi-process Agent Teams.
  • Media Studio: Built-in tools for generating images and video from text or images across multiple providers.
  • Knowledge Management: Project and user-specific wikis (KMS) and a background process (/dream) to audit and consolidate sessions.
  • Browser Automation: Ability to drive a real Chrome session as a tool for navigation and and form filling.
  • Enterprise Ready: Signed organizational policies for branding, OIDC SSO, and tool-call audit trails.

Related

  • Project
    microclaw/microclaw

    A self-hosted Rust agent platform providing a durable agent engine that can be deployed as a server, a native desktop app, or embedded via an SDK.

  • Project
    Superagentsys/novalclaw

    OmniNova is a local-first enterprise security agent that allows users to run AI agents on their own machines to keep model keys and data private while controlling system tools.

  • Project
    zeroclaw-labs/zeroclaw

    A self-hosted agent runtime written in Rust that lets you run a private AI assistant with local control over data, providers, and hardware interaction.

  • Project
    SAIL-Research-Lab/cheetahclaws

    A Python-native agent harness for AI coding assistants that supports any LLM, including local models, providing a flexible and open-source alternative to proprietary tools.

  • Project
    ultraworkers/claw-code

    A Rust-based CLI agent harness that allows users to run AI coding agents locally with support for multiple LLM providers.