Dash MCP Server
# mcp-server-dash
A Model Context Protocol (MCP) server that provides tools to interact with the [Dash](https://kapeli.com/dash) documentation browser API.
Dash 8 is required. You can download Dash 8 at https://blog.kapeli.com/dash-8.
mcp-name: io.github.Kapeli/dash-mcp-server
<a href="https://glama.ai/mcp/servers/@Kapeli/dash-mcp-server">
<img width="380" height="200" src="https://glama.ai/mcp/servers/@Kapeli/dash-mcp-server/badge" alt="Dash Server MCP server" />
</a>
## Overview
The Dash MCP server provides tools for accessing and searching documentation directly from Dash, the macOS documentation browser. MCP clients can:
- List installed docsets
- Search across docsets and code snippets
- Load documentation pages from search results
- Enable full-text search for specific docsets
### Notice
This is a work in progress. Any suggestions are welcome!
## Tools
1. **list_installed_docsets**
- Lists all installed documentation sets in Dash
2. **search_documentation**
- Searches across docsets and snippets
3. **load_documentation_page**
- Loads a documentation page from a `load_url` returned by `search_documentation`
4. **enable_docset_fts**
- Enables full-text search for a specific docset
## Requirements
- macOS (required for Dash app)
- [Dash](https://kapeli.com/dash) installed
- Python 3.12 or higher
- uv
## Configuration
### Using uvx
```bash
brew install uv
```
#### in `claude_desktop_config.json`
```json
{
"mcpServers": {
"dash-api": {
"command": "uvx",
"args": [
"--from",
"git+https://github.com/Kapeli/dash-mcp-server.git",
"dash-mcp-server"
]
}
}
}
```
#### in `Claude Code`
```bash
claude mcp add dash-api -- uvx --from "git+https://github.com/Kapeli/dash-mcp-server.git" "dash-mcp-server"
```
TDQS
Scored across 4 tools
Each tool has a clearly distinct purpose with no overlap. enable_docset_fts configures search capabilities, list_installed_docsets enumerates available resources, search_documentation finds content, and load_documentation_page retrieves specific pages. The boundaries between these operations are well-defined and unambiguous.
All tool names follow a consistent verb_noun pattern with clear, descriptive action-object pairs. The naming convention is uniform throughout: enable_docset_fts, list_installed_docsets, load_documentation_page, and search_documentation all maintain the same structure and style.
Four tools is reasonable for a documentation search server, covering core operations of listing, searching, and viewing documentation. While this is a minimal but functional set, it could potentially benefit from additional utilities like docset management or search refinement tools, but the current count is appropriate for the basic scope.
The toolset provides complete coverage for the core documentation search workflow: discovering available docsets, enabling search features, searching across documentation, and loading specific pages. The only minor gap is the absence of docset installation/removal tools, but the server appears focused on search operations rather than full docset management.