> ## Documentation Index
> Fetch the complete documentation index at: https://docs.2extract.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Overview

> Machine-readable forms of the 2extract documentation - text files following the llms.txt standard, per-page markdown, and a hosted docs MCP server.

Everything in this section exists for one reason: a language model reading a documentation site built for humans wastes tokens on navigation, and often reads the wrong thing.

So we publish the same content in three shapes a model can consume directly. All three are free, need no key, and are read-only.

<CardGroup cols={2}>
  <Card title="LLM References" icon="file-lines" href="/ai/for-agents/llm-references">
    Two text files with the whole portal in them, following the llms.txt standard.
  </Card>

  <Card title="Docs MCP Server" icon="magnifying-glass" href="/ai/for-agents/docs-mcp-server">
    Live search over this documentation, for MCP-capable clients.
  </Card>
</CardGroup>

***

## Which one fits your setup

| Your situation                                      | Use                                                                      |
| :-------------------------------------------------- | :----------------------------------------------------------------------- |
| Building an integration with an AI coding assistant | [llms-full.txt](/ai/for-agents/llm-references), loaded once into context |
| Agent has web access and can fetch what it needs    | [llms.txt](/ai/for-agents/llm-references) as an index                    |
| Client supports MCP                                 | [Docs MCP Server](/ai/for-agents/docs-mcp-server)                        |
| You want one specific page as clean text            | Append `.md` to any URL on this site                                     |

That last one is worth knowing on its own:

```
https://docs.2extract.com/proxy-products/configuration/geo-targeting.md
```

Same page, no navigation, no HTML.

***

## Not to be confused with the MCP Server

There are two MCP servers in this documentation and they do completely different things.

|                         | [MCP Server](/ai/mcp-server/overview) | [Docs MCP Server](/ai/for-agents/docs-mcp-server) |
| :---------------------- | :------------------------------------ | :------------------------------------------------ |
| Endpoint                | `mcp.2extract.com/mcp`                | `docs.2extract.com/mcp`                           |
| What it does            | Creates and manages proxies           | Searches this documentation                       |
| Needs a key             | Yes                                   | No                                                |
| Costs money             | The proxies it creates do             | No                                                |
| Can change your account | Yes                                   | No                                                |

If you want an agent that *does* things - provisions proxies, sets limits, reads your balance - you want the [MCP Server](/ai/mcp-server/overview). This section is about an agent that *knows* things.

***

## Using both together

They complement each other well. A coding assistant with the documentation loaded writes correct usernames and picks the right session parameters on the first attempt, because it knows the format. The same assistant connected to the MCP server can then create the proxy it just wrote code for.

<Tip>
  If you only do one thing: point your coding assistant at `llms-full.txt` before you start building the integration. Most of the time people lose to proxy integration is spent on the username parameter format, which is all in there.
</Tip>

***

## Next steps

<CardGroup cols={2}>
  <Card title="LLM References" icon="file-lines" href="/ai/for-agents/llm-references">
    The two files, what is in each, and how to load them.
  </Card>

  <Card title="Docs MCP Server" icon="magnifying-glass" href="/ai/for-agents/docs-mcp-server">
    Connect a client to live documentation search.
  </Card>
</CardGroup>