# Skills

There is no official June skill yet. You don't need one to give an agent accurate June context: use the [MCP server](https://june.dev/docs/agents/mcp.md) when your client supports it, or [`llms.txt`](https://june.dev/llms.txt) and the [Markdown docs](https://june.dev/docs/agents/markdown.md) otherwise.

## What a skill would add

The docs say what June does, and the MCP server lets an agent search and read them as it works. A skill would add the procedure: what to do, in what order, as instructions an agent loads when the task calls for it. It would point into the docs rather than repeat them:

- **When to use June.** Several model calls that depend on each other, or that mix with your own async code, are June's job; one call or one agent loop is the AI SDK's. [What June doesn't do](https://june.dev/docs/workflows.md#what-june-doesnt-do) is the boundary.
- **How to write a workflow.** A file that default exports a component, step ids derived from the data they're about, and `<Parallel>` around independent work. [Step identity](https://june.dev/docs/workflows/step-identity.md) and [Parallelism and concurrency](https://june.dev/docs/workflows/parallelism.md) have the rules.
- **What to run to check it.** `june doctor` for keys, `june run --json` for output a script can read, and tests with a mock model from `ai/test`, so they need no key or network. See [CLI](https://june.dev/docs/cli.md) and [Testing](https://june.dev/docs/workflows.md#testing).
- **Where to look things up.** The MCP server's `search_docs` and `get_doc`, or `llms.txt`.

When a June skill ships, this page will say how to install and use it.

## Next

- [MCP server](https://june.dev/docs/agents/mcp.md): connect a client.
- [Quickstart](https://june.dev/docs/quickstart.md): the same procedure, for people.