GitAgent: Open standard defining AI agents as files in Git repositories.

What it does

GitAgent is an open specification for defining AI agents as files stored in Git repositories. The standard uses three core files: agent.yaml for configuration, SOUL.md for personality and instructions, and SKILL.md for capabilities. This file-based structure allows agents defined once to be exported to multiple frameworks including Claude Code, OpenAI Agents SDK, CrewAI, Google ADK, and LangChain.

Being Git-native means agent behavior changes can be version controlled, rolled back, and tracked like code changes. The spec treats agents as portable artifacts rather than framework-locked configurations.

Who it is for

Developers building AI agents across multiple frameworks benefit from avoiding rewrite work when switching tools. Teams managing multiple agents find value in standardized definitions and version history. Organizations wanting framework flexibility without vendor lock-in are the primary audience.

Pricing

The site does not show prices.

How it stands out

Most agent frameworks define their own proprietary formats, forcing rewrites when switching. GitAgent sidesteps this by proposing a framework-agnostic standard. The Git-native approach provides version control, collaboration features, and audit trails built-in without additional tooling. The spec appears to target portability as its core differentiator rather than adding new capabilities on top of existing frameworks.

What a founder should check

First, verify adoption momentum among the target frameworks. GitAgent only solves the switching problem if enough frameworks support the standard; without adoption from OpenAI, Anthropic, and others, the portability claim remains theoretical.

Second, test the real switching costs in practice. Does exporting from GitAgent to CrewAI actually require zero rewrites, or do framework-specific extensions force manual work? Check whether the spec covers enough agent functionality to be useful versus covering only basic cases.

Third, assess whether this captures the actual pain point. Developers might avoid framework switching more because of integration investments in databases, APIs, and monitoring rather than agent definition syntax. If the real cost lies elsewhere, a spec solves only part of the problem.

Thinking of building something like this?

Every launch here is a competitor to somebody's idea. If yours is close, check it against the market before you build: the Full Check names the rivals, the prices and the gaps.