Tokitoki — Coding Time & AI Usage Analytics

Know where your time and your AI spend actually go.

Tokitoki automatically tracks your coding activity in VS Code and syncs your local AI agent usage — Claude Code, Codex, Copilot, Gemini, and a dozen more — into one dashboard at tokitoki.dev.

Features

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.

Quick Start

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.

Commands

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: Toggle Tracking`	Enables or disables Tokitoki

Settings

Setting	Default	Description
`tokitoki.enabled`	`true`	Enable tracking and AI usage sync
`tokitoki.autoSync`	`true`	Scan AI usage on startup and every 30 minutes
`tokitoki.statusBar.enabled`	`true`	Show the status bar item
`tokitoki.showNotifications`	`true`	Notify on failures and manual commands
`tokitoki.baseUrl`	(empty)	Self-hosted / staging server override
`tokitoki.logLevel`	`info`	Output channel verbosity

To pin a stable project name across machines and editors, put a .tokitoki file in the project root — the first line is the project name, an optional second line overrides the branch.

Privacy

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.

Tokitoki