> ## Documentation Index
> Fetch the complete documentation index at: https://docs.plungeai.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Docs for AI agents

> Dropped: PlungeAI runs no web crawler, so webmaster guidance does not apply. Replacement: how agents read these docs (`/agents.md`, llms.txt, `.md` twins, Docs MCP).


<Note>
Partly available: the Docs MCP server, which will let your agent search and read these pages. Tracked as TI-35.
</Note>

Every page on docs.plungeai.com has a markdown form, the whole site has an index, and one file tells an agent how to connect PlungeAI.

## /agents.md

[https://docs.plungeai.com/agents.md](/agents.md) is the setup file for coding agents: get an `ozk_` key from the user, connect the client to `https://mcp.plungeai.com/v1`, verify with `plungeai_whoami`, add the skills and follow the rules. Paste this prompt into any agent to run it:

```text
Use curl to read https://docs.plungeai.com/agents.md and perform the setup to connect PlungeAI
```

It is served as `text/markdown; charset=utf-8`, and it is not a sidebar page.

## llms.txt and llms-full.txt

| File | What it holds |
|---|---|
| [/llms.txt](/llms.txt) | `# PlungeAI`, a one-line description, then one bullet per page in sidebar order (`[Title](https://docs.plungeai.com/<page>.md): description`), with ` (planned)` after pages that are not available yet; then `## OpenAPI Specs` and `## Optional` |
| [/llms-full.txt](/llms-full.txt) | Every page in full: `# <title>`, a `Source:` line with the page URL, then the page body; API and MCP reference pages in a condensed form |

## Markdown twins

Append `.md` to any page URL, for example [/getting-started/quickstart.md](/getting-started/quickstart.md), or request the page with `Accept: text/markdown`:

```bash
curl -s -H 'Accept: text/markdown' https://docs.plungeai.com/getting-started/quickstart
```

A twin starts with a pointer to `/llms.txt`, then the page title and description, then the page source unchanged. Twins carry `X-Robots-Tag: noindex`, so search engines index the HTML page only.

## Docs MCP

A docs MCP server at `docs.plungeai.com/mcp` will let an agent search these docs and read a page by path. It is planned (TI-35) and answers with a JSON-RPC error until then. The product MCP server at `https://mcp.plungeai.com/v1` is a different server and is live.

<Card title="Docs MCP server" icon="plug" href="/developer-tools/mcp/docs-mcp">
What the docs MCP server will do.
</Card>

## Other llms.txt files

| URL | Covers |
|---|---|
| https://api.plungeai.com/llms.txt | The One API guide |
| https://mcp.plungeai.com/llms.txt | Workflow authoring over MCP (the CNL skill) |
| https://skills.plungeai.com/llms.txt | The agent skills catalog |