Tokitoki automatically tracks your coding activity in VS Code and syncs your AI agent usage — Claude Code, Copilot, Gemini, and a dozen more — into one dashboard at tokitoki.dev.
- Automatic time tracking — just code. Tokitoki records which files, projects, and languages you work in, including debugging and build time. No timers to start, no forms to fill.
- AI usage analytics — your local AI coding agents (Claude Code, Codex, Copilot CLI, Gemini, Amp, Goose, OpenCode, and more) are scanned and synced automatically, so tokens, models, and costs show up next to your coding time.
- One dashboard for everything — daily timelines, per-project totals, language breakdowns, and AI cost tracking across every editor and machine you use.
- Works offline — activity is queued locally and uploaded when you are back online. Nothing is lost on a plane.
- Open source — the extension and its CLI are open source, and the same tracking works in JetBrains IDEs, Eclipse, and the macOS menu bar app.
- Install the extension.
- Run Tokitoki: Set API Key from the command palette and paste the key from tokitoki.dev — the extension will also prompt you on first use.
- That's it. Click the status bar clock anytime to open your dashboard, already signed in.
| Command | What it does |
|---|---|
Tokitoki: Set API Key |
Stores your API key for every Tokitoki client |
Tokitoki: Open Dashboard |
Opens your web dashboard, signed in |
Tokitoki: Show API Key Status |
Shows the configured key, masked |
Tokitoki: Set Project Name |
Pins the name this folder reports, in .tokitoki |
tokitoki.statusBar.enabled (default true) shows the status bar item.
Tracking and uploading need no configuration — they start with the editor and run for the session.
To pin a stable project name across machines and editors, run Tokitoki: Set
Project Name — it writes a .tokitoki file in the folder root, pre-filled
with the name currently in use (the folder name, until you change it). The
first line of that file is the project name, an optional second line overrides
the branch; editing it by hand works just as well, and the next heartbeat
picks it up.
Tokitoki records activity metadata only: file paths, project and branch
names, language, timestamps, and cursor position — never your code. AI
usage sync reads token counts and model names from your local agent data.
Everything is queued in ~/.tokitoki and uploaded over HTTPS with your API
key; delete your data anytime from the dashboard.