Time tracking for developers Hours per repo, including the time your AI agents put in.

A time tracker built around how developers actually work

A developer’s day moves between an editor, a terminal, pull requests in the browser, and now one or more AI agents running alongside all of it. A time tracker that only watches the IDE misses most of that. TimeUno runs on the whole computer, records the app or site in front of you minus idle time, and reads which project you’re in from your editor or AI coding tool, so the hours land under the right repo without a timer.

Every repo becomes a project, without setup

TimeUno reads the project name from package.json, Cargo.toml, go.mod or pyproject.toml, or else the folder name, and asks before it tracks a new one.

1. Open a repo in your editor or AI coding tool. The desktop widget asks Track X as a project?
2. Choose Add, and time in that repo files there from then on.
3. For work that isn’t in a repo, add a project by hand on the Projects page, with keywords that match your window titles or URLs.
4. Move anything that landed in the wrong place from the Timeline; TimeUno remembers the correction.

Your AI coding agents, tracked next to you

The desktop app finds these tools without a plugin and records whether each is working or waiting for you:

• Claude Code
• Cursor
• Codex CLI
• OpenCode
• GitHub Copilot
• Windsurf
• Aider
• Ollama
• Gemini CLI

AI time sits beside your own tracked hours on the same project, not added to them, so a repo with an agent running for an hour while you’re in a meeting still shows the hours you actually worked, split from the agent’s.

Agents on a remote server

If you run Claude Code or Cursor CLI over SSH, the desktop app on your laptop can’t see those sessions. The headless agent can:

curl -fsSL https://app.timeuno.io/install-agent.sh | bash

It installs as a background service, tracks Claude Code, Cursor CLI, Codex CLI, Gemini CLI, OpenCode, Aider and Ollama, and tracks each running session separately, with its project and whether it’s working or waiting for you. Two agents on the same server show up as two sessions, not one.

Ask Claude or Cursor about your own tracked time

TimeUno has an MCP server at https://app.timeuno.io/mcp. Connect it from Claude, ChatGPT, Claude Code or Cursor, sign in, and ask questions such as “What did I work on this week?” or “Which repo did the AI agents spend the most time on?”. It can read your summary, timeline, current activity and project time, and, if you allow it, pause tracking or add a note. It never gets screenshots or on-screen text.

What never leaves your device

Sends	Never leaves your device
When an AI tool session starts and ends	Your prompts
Whether it’s working or waiting for you	The AI’s answers
The model, token count and cost	Your code
The project folder and conversation title

Not every tool shares all of that: Aider, Windsurf and Ollama only report whether they’re active, so they have no token count or cost. TimeUno runs on servers at Hetzner in Germany, and nothing is sold or used for advertising.

Updated 18 October 2026