MCP Server - Connect an AI assistant
Clocktopus exposes a remote Model Context Protocol (MCP) server, so any MCP-compatible AI assistant can read your tracked hours on your behalf; It is read-only and scoped to your own data.
What you need
- A Clocktopus account (you sign in during the connection flow).
- An MCP client that supports remote servers over OAuth (most AI assistants and agent tools that offer “custom connectors” do).
- The server URL:Bash
https://clocktopus.app/api/mcp
No API keys. This functionality is powered by secure OAuth 2.1 authorization, which ensures all actions respect your existing access controls. The client walks you through a standard sign-in and a one-time consent screen, the same login you use for the dashboard, so a connected assistant can only ever see the data you can. Nothing to copy or paste.
Add the connector
The exact wording varies by client, but the flow is the same everywhere.
Step 1: Add a custom / remote MCP server
In your client's connector or integration settings, choose to add a custom (remote) MCP server.
Step 2: Enter the server URL
https://clocktopus.app/api/mcpLeave any optional OAuth Client ID / Secret fields blank, the client registers itself automatically.
Step 3: Sign in & approve
A browser window opens to Clocktopus. Sign in (if you aren't already), then approve the access request on the consent screen. The client finishes connecting automatically.
Step 4: Use it
Ask your assistant something like “Using Clocktopus, list my projects” or “how many hours did I track per task on Public Website in May?” It calls the relevant Clocktopus tools and answers from your data.
Good to know
- Read-only & scoped to you. The connector can only read time data your account already has access to — it can't create, edit, or delete anything.
- Revoke any time. Remove the connector in your MCP client to stop access; the issued token simply expires.
- Date format. Dates are
YYYY-MM-DD. Double-check the year — a mismatched range is the usual reason totals look off.