MCP server
GitDash speaks the Model Context Protocol, so an AI assistant can look things up in these docs while it helps you: setup, configuration, access control and what every metric means. Add /mcp on any GitDash host as a remote MCP server. It is read-only and needs no sign-in.
Connect
| Client | How |
|---|---|
| Claude (claude.ai, Desktop) | Settings → Connectors → Add custom connector, URL https://www.gitdash.info/mcp |
| Claude Code | claude mcp add --transport http gitdash https://www.gitdash.info/mcp |
| Cursor | Add the server to .cursor/mcp.json (below) |
| MCP Inspector | npx @modelcontextprotocol/inspector |
.cursor/mcp.json
{
"mcpServers": {
"gitdash": { "url": "https://www.gitdash.info/mcp" }
}
}A self-hosted GitDash serves the same endpoint on its own domain.
Tools
| Tool | What it returns |
|---|---|
list_docs | Every docs page with its id, title, summary and URL. |
search_docs(query) | The five best-matching pages, each with an excerpt and URL. |
get_doc(page) | One whole page as Markdown, by id. |
explain_metric(metric) | What a metric measures, how it is calculated and what good looks like — e.g. lead time, change failure rate, MTTR, bus factor. |
Try asking: “Using GitDash, what does change failure rate measure?” or “How do I deploy GitDash with Helm?”
Limits
The endpoint allows 300 requests a minute per IP address on each server instance. Tools only read public documentation; your dashboards and GitHub data are not reachable through it.