# EventCatalog Agents

Copy as Markdown[View as Markdown](/docs/development/ask-your-architecture/agents/overview.md)

***

EventCatalog Agents are AI agents that help you manage and document your architecture with EventCatalog. They are open and available on GitHub at [event-catalog/agents](https://github.com/event-catalog/agents).

They understand EventCatalog conventions (services, events, commands, queries, domains, channels, containers, specifications, and more) and use that understanding to keep your architecture documented for you, without you having to hand-write and maintain every catalog file yourself.

## How EventCatalog Agents work[​](#how-eventcatalog-agents-work "Direct link to How EventCatalog Agents work")

EventCatalog Agents are powered by an AI model and guided by the same [Skills](/docs/development/ask-your-architecture/skills/introduction.md) that teach any AI agent how to produce correct EventCatalog documentation. This means an agent doesn't just guess at documentation. It follows EventCatalog's conventions for frontmatter, folder structure, and resource relationships.

At a high level, an agent:

1. **Looks at a change** in your system, for example the diff of a pull request that adds a new event or service.
2. **Plans the impact**. It works out whether the change requires any documentation updates, and if so, exactly which catalog resources should be created or updated.
3. **Applies the plan**. It writes the documentation, following EventCatalog conventions, and validates its own work.
4. **Proposes the changes**. Rather than editing your catalog silently, it opens a pull request you can review, approve, and merge.

The deterministic parts (reading diffs, resolving paths, opening pull requests) are handled in code. The agent is responsible only for reasoning about what your documentation should say.

## Where agents run[​](#where-agents-run "Direct link to Where agents run")

Today, EventCatalog Agents run in your **CI/CD pipeline as a GitHub Action**. Each time you open a pull request, the agent reviews the change and proposes the matching documentation updates in a separate pull request against your catalog. CI is the first place we run them. It is a delivery surface, not the limit of what the agents are.

Early access

EventCatalog Agents are in early access. They are free for anyone to evaluate today. In the future, a license will be required to run EventCatalog Agents in production.

## Your infrastructure, your model, your data[​](#your-infrastructure-your-model-your-data "Direct link to Your infrastructure, your model, your data")

EventCatalog Agents are designed so that you stay in control of your code, your models, and your data.

* **They run on your own infrastructure.** The agent runs inside your own CI/CD pipeline (for example, GitHub Actions), not on EventCatalog servers. Your source code is read where it already lives and never leaves your environment to reach us.
* **You bring your own model.** You choose the model and provide your own provider API key (such as Anthropic, OpenAI, or OpenRouter). The agent talks directly to the provider you configured, using your account. You can switch providers or models at any time.
* **Your data stays yours.** Because the agent runs in your pipeline and calls your model provider directly, your code and documentation are processed under your own infrastructure and your own provider agreement. EventCatalog does not receive or store the contents of your repositories.

This means you can adopt EventCatalog Agents without handing your architecture or source code to a third party.

## Available agents[​](#available-agents "Direct link to Available agents")

| Agent                                                                          | What it does                                                                                                           |
| ------------------------------------------------------------------------------ | ---------------------------------------------------------------------------------------------------------------------- |
| [Code-to-Docs](/docs/development/ask-your-architecture/agents/code-to-docs.md) | Keeps your EventCatalog documentation in sync with your code by reviewing pull requests and proposing catalog updates. |

## Have an idea for an agent?[​](#have-an-idea-for-an-agent "Direct link to Have an idea for an agent?")

More agents and AI workflows are on the way. If you have an idea for an agent or an AI workflow you'd like to see in EventCatalog, [open an issue on GitHub](https://github.com/event-catalog/agents/issues/new) and let us know. We use these ideas to decide what to build next.