Timebook
Time tracking, invoicing, and lightweight bookkeeping for solo LLCs and freelancers. Timebook ships with an MCP server so AI agents (Claude Code, Cursor, Claude.ai, any MCP-speaking client) can manage your time directly — start a timer when you begin a task, log the entry when you're done, or just ask "log 45 min on the Acme refactor."
No browser tab. No "did I forget to start the timer?"
What you can do from your agent
- Start / stop timers on any project
- Log a discrete time entry with a duration ("log 1h 15m on the API refactor")
- List clients, projects, and active timers
- Query recent entries
- Inspect / update / delete a specific entry
Install — stdio (most common)
For Claude Desktop, Cursor, Claude Code, Continue, or any MCP client that supports stdio:
```json { "mcpServers": { "timebook": { "command": "npx", "args": ["-y", "@squidcode/timebook", "mcp"] } } } ```
Then in a terminal, run once:
```bash npx @squidcode/timebook login ```
That opens a browser, logs you into Timebook, and writes a scoped token to a local config file (mode
0600). The MCP server reuses that token — no token paste required.
Install — remote / Claude.ai web
Timebook also runs as a hosted Streamable-HTTP MCP server with full OAuth 2.0:
- Endpoint:
https://usetimebook.com/mcp - Auth-server metadata:
https://usetimebook.com/.well-known/oauth-authorization-server
In Claude.ai → Settings → Connectors → Add → paste https://usetimebook.com/mcp. You'll see Timebook's
consent screen once, then Claude has access to the tools below.
Tools exposed
| Tool | What it does |
|---|---|
whoami | Current authenticated user |
list_clients | All clients in scope |
list_projects | All projects in scope |
get_active_timer | The running timer, or null |
start_timer | Start a timer on a project |
stop_timer | Stop the active timer |
log_time | Log a complete entry with duration (preferred for AI-driven entries) |
list_entries | Recent entries with filters |
update_entry | Edit an existing entry |
delete_entry | Remove an entry |
Why an MCP server for time tracking?
Time tracking is uniquely well-suited to agentic workflows. Your agent already knows when you started a task, what file you were in, when you wrapped up. Existing trackers assume a human flipping a switch in a browser tab — Timebook flips it around: your AI logs the time, you stay in flow.
Links
- Web: https://usetimebook.com
- CLI + MCP repo: https://github.com/squidcode/timebook-cli
- npm: https://www.npmjs.com/package/@squidcode/timebook
Built for solo LLCs and freelancers. One price. No tiers, no per-seat, no upsells.
服务器配置
{
"mcpServers": {
"timebook": {
"command": "npx",
"args": [
"-y",
"@squidcode/timebook",
"mcp"
]
}
}
}