# Documentation Search MCP server

> Provides local RAG capabilities for markdown documentation by retrieving grounded passages and citations from local paths or git repositories.

- Repository: https://github.com/alyiox/mcp-docs-ask
- Page: https://hostedmcp.org/servers/alyiox/mcp-docs-ask
- Transport: stdio
- Stars: 0
- License: MIT
- Latest version: 0.4.0

## Install

### Claude Code

```sh
claude mcp add mcp-docs-ask -- uvx mcp-docs-ask
```

### Cursor (`~/.cursor/mcp.json`)

```json
{
  "mcpServers": {
    "mcp-docs-ask": {
      "command": "uvx",
      "args": [
        "mcp-docs-ask"
      ]
    }
  }
}
```

### VS Code (`.vscode/mcp.json`)

```json
{
  "servers": {
    "mcp-docs-ask": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "mcp-docs-ask"
      ]
    }
  }
}
```

### VS Code CLI

```sh
code --add-mcp '{"name":"mcp-docs-ask","type":"stdio","command":"uvx","args":["mcp-docs-ask"]}'
```

### Claude Desktop (`claude_desktop_config.json`)

```json
{
  "mcpServers": {
    "mcp-docs-ask": {
      "command": "uvx",
      "args": [
        "mcp-docs-ask"
      ]
    }
  }
}
```

### Codex

```sh
codex mcp add mcp-docs-ask -- uvx mcp-docs-ask
```

### Codex config.toml (`~/.codex/config.toml`)

```toml
[mcp_servers.mcp-docs-ask]
command = "uvx"
args = ["mcp-docs-ask"]
```

### Gemini CLI

```sh
gemini mcp add -s user mcp-docs-ask uvx -- mcp-docs-ask
```

### Gemini settings.json (`~/.gemini/settings.json`)

```json
{
  "mcpServers": {
    "mcp-docs-ask": {
      "command": "uvx",
      "args": [
        "mcp-docs-ask"
      ]
    }
  }
}
```

## Tools (3)

- `list_docs`
- `ask_docs`
- `reindex`

## FAQ

### Does this server provide LLM answers?

No, the server performs retrieval only. The MCP host, such as Cursor or Claude, is responsible for synthesizing the final answer from the retrieved passages.

### Can I use this with private git repositories?

Yes, you can use private git URLs as long as git credentials (like SSH or HTTPS helpers) are configured on your machine.

### Do I need to manually rebuild the index when I change my local files?

No, the index automatically refreshes itself when local files, configuration, or the embedding model change.

### What embedding models are supported?

Any Hugging Face ID loadable by sentence-transformers works, including models optimized for English-only, Chinese-only, or multilingual use.
