Connect to Claude (MCP)
Connecting a local AI client takes two things: a workspace API key and one command in your project folder. Both are handled in Settings → Connect to Claude.
1. Generate a workspace API key
In Settings → Connect to Claude, generate an API key. The key (prefixed xpt_…) is shown once — copy and store it. It identifies your workspace and you as the acting user, so anything the connected assistant does is attributed to you. You can manage and revoke keys in the API Keys panel; create a separate key per machine or client.
This is different from an agent's provider key (OpenAI/DeepSeek/…). The workspace key lets an external client connect into the workspace; the agent key lets an in-workspace agent call an LLM.
2. Add the MCP server
Wire the MCP server in each project folder with that folder's API token. Run inside the project root (the panel shows this command with your key already inlined):
claude mcp add --scope project --transport http xpt-tracker <MCP_URL> --header "Authorization: Bearer <your-key>"
Per-folder wiring means different projects can talk to different XPT workspaces with different keys. For other MCP clients (Cursor, Codex, …), drop the same connection details into the client's MCP config:
{
"mcpServers": {
"xpt-tracker": {
"type": "http",
"url": "<MCP_URL>",
"headers": { "Authorization": "Bearer <your-key>" }
}
}
}
Use the exact, copy-paste values from the panel — they're filled in for your deployment.
What the assistant can do — the toolset
Tools are named domain_action, so they group by area. At a high level:
| Area | What the tools cover |
|---|---|
| Workspace & people | Who am I, workspace overview, member list |
| Teams | Teams and their states, labels, priorities, members |
| Tasks | List, get, create, update, batch-update, archive, and comment on tasks |
| Projects | List, get, create, update, archive; states, status updates, dependencies |
| Milestones | List, get, create, update, archive |
| Knowledge | Browse the tree, read, search, create, update, delete pages |
| Channels | List, get, create, archive, and post (mentions can spawn agents) |
| Brainstorms | List, get, search, create, and post |
| AI Workflows | Your workspace's processes as runnable tools (wf_task_plan, …) — plus reading and authoring the workflows themselves. See "Workflows in Your AI Client". |
So a connected assistant can run essentially the whole workspace, not just read it — and it follows your workspace's processes while doing so.
Safety
- Every tool call goes through the same permission checks as the web app — the assistant only sees and changes what your key allows.
- Mutating tools use an explicit "changes" model, so the assistant can't blank a field by leaving it out.
- Actions appear in the relevant Activity feed, attributed to you.
- Revoke a key any time from the API Keys panel to cut off a client.