Xenmark MCP server
Official# Xenmark MCP server
An [MCP](https://modelcontextprotocol.io) server that lets AI assistants — Claude, Cursor, VS Code and others — read your [Xenmark](https://www.xenmark.app) drawing-review data: projects, drawings, revisions, comments, replies and your notification inbox. Read-only, runs on your own machine, uses a token you create inside Xenmark.
**Status:** released. On [npm](https://www.npmjs.com/package/xenmark-mcp) as `xenmark-mcp` and in the [official MCP Registry](https://registry.modelcontextprotocol.io) as `io.github.Xenmark/xenmark-mcp`.
## What you can ask
- "Anything new for me in Xenmark?" / "Who mentioned me today?"
- "Which projects am I on, and what's my role?"
- "What's still open on drawing 1001-300, Rev. B?"
- "List the unresolved comments on that revision and who raised them."
- "What did Greta reply on comment #7?"
Every answer carries a link straight into Xenmark.
## What it never does
It never writes to Xenmark, never sees your password, never returns e-mail addresses, files, billing or account roles, and never reaches projects you are not a member of. It talks to exactly one server: `https://api.xenmark.app`, the official read-only Integration API ([contract](https://github.com/Xenmark/xenmark-api)).
## Get a token
In Xenmark: profile → **Account & Security → Connected apps → Connect an app → "Personal API token (Postman, GitHub)" → Generate**. The token is shown once; keep it in the assistant's config only. Revoke it any time from Connected apps.
## Install
Requires Node.js 18 or newer. The server runs with `npx -y xenmark-mcp` — nothing to clone or build. (To run from a source checkout instead: `npm install && npm run build`, then point your assistant at `dist/index.js`.)
**Claude Desktop** — edit `claude_desktop_config.json` (Settings → Developer → Edit Config) and add:
```json
{
"mcpServers": {
"xenmark": {
"command": "npx",
"args": ["-y", "xenmark-mcp"],
"env": { "XENMARK_TOKEN": "xmk_live_YOUR_TOKEN_HERE" }
}
}
}
```
Restart Claude Desktop; the Xenmark tools appear under the tools icon.
**Cursor** — Settings → MCP → Add new server, same `command`, `args` and `env`.
**VS Code** (Copilot agent mode) — add the same entry under `"servers"` in `.vscode/mcp.json` or your user `mcp.json`.
From a source checkout, replace the command with `node` and the args with the full path to `dist/index.js`.
## Support
support@xenmark.app · Security: see [SECURITY.md](SECURITY.md). Never post a token anywhere public.
TDQS
Scored across 9 tools
Each tool targets a distinct resource or action: identity, notifications, projects, drawings, revisions, comment rollup, comments, replies, and drawing search. The only potential overlap (list_drawings vs. find_drawing) is explicitly clarified by descriptions, so an agent can reliably choose the right tool.
All names use a consistent snake_case style with a uniform 'xenmark_' prefix, which makes them predictable and readable. However, the semantic pattern is slightly mixed: most follow list_<noun>, but there are also find_drawing, review_summary, recent_activity, and whoami, so it is not a strict verb_noun convention throughout.
Nine tools is well-scoped for a read-only data-access server covering projects, drawings, revisions, comments, and activity. Each tool earns its place without redundancy, and the set avoids the bloat that would come from exposing every possible field or operation separately.
The read surface is thorough: identity, activity, projects, drawings, revisions, comments, replies, and a review summary are all covered. However, write operations that are central to a review workflow—posting comments/replies, changing comment status, or marking activity read—are entirely absent, creating a notable lifecycle gap.