AI Assistants (MCP)

aTimeLogger has an official open-source MCP (Model Context Protocol) server that connects AI assistants — Claude Desktop, Claude Code, ChatGPT, Cursor and any other MCP client — to your account. Once connected, you can start and stop timers, log past activities, edit comments and tags, and get time reports just by asking your assistant in plain language. The server can also run on a host you control as a remote connector, which makes it available in claude.ai in the browser, the Claude mobile apps and ChatGPT (Developer Mode).

You will need a personal access token (Settings → API Tokens in the web app; API access is part of the Premium Sync plan). The server is published on npm, so no download or build step is needed.

Setup instructions, configuration snippets and the full tool list are on GitHub: github.com/zaplitny/atimelogger-mcp. For an overview of what the integration can do, see atimelogger.pro/mcp.html.

Example conversation

You: Start tracking Work — I actually began at 11:30

Claude: Done — Work is running, backdated to 11:30.

You: I also read for 2 hours yesterday before midnight, log it

Claude: Logged — Reading, yesterday 22:00–00:00 (2h).

You: Add a comment to it: finished chapter 5

Claude: Updated — comment added to yesterday's Reading entry.

You: Where did my week go?

Claude: You tracked 73h 20m this week: Sleep 38h 10m, Work 24h 45m, Reading 8h 30m, Sport 3h 55m. Sleep and Work are on pace with last week; Reading is up 2 hours.

Activity names are matched automatically ("start dev" finds your "Development" type), and the assistant chains actions on its own — "stop whatever is running and start Work" is a single request.