open-gitagent/gitagent

A universal git-native AI agent framework. Your agent lives inside a git repo — identity, rules, memory, tools, and skills are all version-controlled files.

What it solves

Gitagent is a framework for creating AI agents where the agent's entire state—including its identity, rules, memory, and tools—is stored as version-controlled files within a git repository. This "agents as repos" approach allows users to fork, branch, and diff the personality and memory of an agent just as they would with source code.

How it works

The agent is defined by a specific directory structure containing several key files:

  • agent.yaml: Configures the LLM model, tools, and runtime settings.
  • SOUL.md and RULES.md: Define the agent's personality and behavioral constraints.
  • memory/: Stores git-committed memory with a full history.
  • tools/ and skills/: Contain declarative tool definitions and composable instruction modules.

Gitagent can be run via a CLI, a programmatic TypeScript SDK, or a voice UI. It also acts as an MCP (Model Context Protocol) client, allowing it to automatically discover and use tools from any MCP server.

Who it’s for

  • Developers who want to version-control their AI agent's configuration and behavioral evolution.
  • Teams looking to share and collaborate on agent personalities and skill sets via git.
  • Users who want a multimodal agent that can interact with local files, shell commands, and external MCP servers.

Highlights

  • Git-Native State: Identity, memory, and rules are all version-controlled files.
  • MCP Integration: Native support for any MCP server for instant tool expansion.
  • Extensible Architecture: Supports plugins, custom declarative tools, and programmatic lifecycle hooks.
  • Flexible Deployment: Offers a slim CLI, a full SDK for in-process integration, and a voice-enabled web UI.

Related

  • Project
    open-gitagent/opengap

    OpenGAP is a git-native, framework-agnostic standard for defining AI agents, allowing them to be versioned, shared, and deployed across different AI platforms using a standardized repository structure.

  • Project
    DotHarness/dotcraft

    An open-source, self-hosted AI agent runtime built with C#/.NET that can be embedded into applications or used as a desktop app to manage extensible AI agents.

  • Project
    matrixorigin/memoria

    Memoria adds Git‑like version control to AI‑agent memory, letting agents store, branch, merge, and roll back facts, preferences, and procedural steps. It provides hybrid vector/full‑text search, automatic contradiction detection, and full audit trails, and can run as a cloud service or self‑hosted with MatrixOne’s copy‑on‑write database.

  • Dispatch
    Version Control for AI Agents: Beyond the Git Commit

    An exploration of Regent VCS, a tool designed to track AI agent intent and session history, and the debate over whether agents need their own version control or can simply use Git.

  • Project
    corespeed-io/zypher-agent

    Zypher Agent is a framework for building production-ready AI agents that use reactive reasoning loops to dynamically execute tasks within applications.