# Docs MCP

> Let Claude, Cursor, VS Code or any MCP client search and read these docs. Free, no API key.

The **Docs MCP** server gives your AI tool these docs: every field guide, concept,
pattern and reference page. Your assistant answers from Skylit's own documentation
instead of from memory.

> **Note:** This is not the [Skylit MCP server](https://www.skylit.ai/docs/mcp/overview). The Docs MCP only reads
> documentation: it is free, needs no API key and uses no credits. The Skylit MCP
> server serves live market data (Flowseeker and Heatseeker), needs an API key and is
> in Beta.

| | |
| --- | --- |
| **Endpoint** | `https://www.skylit.ai/docs/mcp` |
| **Auth** | None |
| **Tools** | `search_docs`, `get_page`, `list_pages` |

## Connect it

### Claude Code

```bash
claude mcp add --transport http skylit-docs https://www.skylit.ai/docs/mcp
```

### Claude Desktop

In **Settings → Connectors → Add custom connector**, paste the endpoint:

```
https://www.skylit.ai/docs/mcp
```

No authentication is needed. (Custom connectors require a Claude plan that
supports them.)

### Cursor

Open any docs page, choose the menu next to **Copy page**, then **Add docs to
Cursor**. Or add it to `~/.cursor/mcp.json` (global) or `.cursor/mcp.json` in a
project:

```json
{
  "mcpServers": {
    "skylit-docs": {
      "url": "https://www.skylit.ai/docs/mcp"
    }
  }
}
```

### VS Code

Open any docs page, choose the menu next to **Copy page**, then **Add docs to
VS Code**. It installs a server named `skylit-docs` of type `http` with the
endpoint above.

## What it can do

| Tool | What it does |
| --- | --- |
| `search_docs` | Finds the best-matching sections for a query, with a snippet and a link to each. Optionally limited to one area (Platform, Heatseeker, Flowseeker, Atlas, Nexus, Tempest, Talon or Developers). |
| `get_page` | Returns one page as Markdown, from a search result's path or URL. |
| `list_pages` | Lists every page with a one-line summary (the same index as [llms.txt](https://www.skylit.ai/docs/llms.txt)). |

Once connected, ask your assistant something like *"Using the Skylit docs, what is a
King Node?"* and it searches the docs and links the pages it used.

## Other ways to read the docs

- Every page is available as Markdown: add `.md` to its URL.
- [llms.txt](https://www.skylit.ai/docs/llms.txt) indexes every page, and
  [llms-full.txt](https://www.skylit.ai/docs/llms-full.txt) has them all in one file.