> **Can't find what you're looking for?** Use `search_docs` on the docs MCP server at `https://sdk.railnet.org/api/mcp` to find what you need.

# AI Agents

This site also serves [`/llms.txt`](https://sdk.railnet.org/llms.txt), [`/llms-full.txt`](https://sdk.railnet.org/llms-full.txt) and every page as
Markdown at the same path with a `.md` suffix. An MCP server at `/api/mcp` searches these docs and
the source on GitHub.

The package ships five [TanStack Intent](https://github.com/TanStack/intent) skills, `railnet-core`,
`railnet-conduit`, `railnet-vehicle`, `railnet-access-control` and `railnet-react`, so an agent reads
the SDK from `node_modules`.

Once per project:

```bash
bunx @tanstack/intent@latest install
```

That writes loading guidance into `AGENTS.md`. If your agent reads `CLAUDE.md` instead, move the
generated `<!-- intent-skills -->` block there; later runs follow it. Swap `bunx` for `pnpm dlx` or
`npx`.

The agent takes it from there:

```bash
bunx @tanstack/intent@latest load @railnetorg/railnet-sdk#railnet-conduit
```

Or hand the whole setup to the agent:

<Prompt value="Install the TanStack Intent skills with `bunx @tanstack/intent@latest install`, then load `@railnetorg/railnet-sdk#railnet-core` and the skill for the module I am working on before writing any Railnet SDK code." />