Connect Time Momentum to your tools.

The command line tm, the agent collector for Claude Code, Codex and Cursor, setup guides your AI agent can follow, and the REST API behind all of it.

  • macOS, Linux and Windows
  • Node.js 22.14 or newer

Let your AI agent set it up.

Paste one sentence into Claude Code, Codex or Cursor. The agent reads the guide, checks Node.js, installs tm and then asks you to connect in a terminal of your own. That is where you enter the API key, so it never appears in the conversation.

Set up the Time Momentum agent collector on this computer. Read and follow this guide: https://www.time-momentum.com/agents/collector.md

Guides for agents in Markdown: agent collector and command line tm. The setup pages in the app show the same prompt.

Capture your AI sessions.

The agent collector records time, model and usage of every Claude Code, Codex and Cursor session through their hooks. Also in Free.

  1. Install tm

    Windows (PowerShell): irm https://app.time-momentum.com/install-cli.ps1 | iex. Then open a new terminal.

    curl -fsSL https://app.time-momentum.com/install-cli.sh | sh
  2. Create an API key

    In the app under Datenquellen › Agent-Kollektor. It is shown once.

  3. Connect

    Asks for the key with hidden input, then which tools to capture and whether to report subscription limits. A rejected key changes nothing.

    tm collect init
  4. Check

    tm collect status

Only measurements leave your computer: time, model, usage, folders and commits. Never source code or file names; of the conversation at most a session title, which you can switch off.

Work with the command line.

A key with read access covers projects and time entries. Everything else needs read and write access, which is part of Pro.

  1. Install tm

    Skip this if the collector already installed it.

    curl -fsSL https://app.time-momentum.com/install-cli.sh | sh
  2. Sign in

    Create the key in the app under Datenquellen › CLI. tm asks for it with hidden input.

    tm auth login
  3. Check login and permission

    tm whoami
  4. Claude Code: install the skill

    tm skill --install
  5. Claude Code: status line

    Project, effort and budget of the current session at the bottom of Claude Code.

    tm statusline install

Or without installing, for example in an agent sandbox: npx -y @time-momentum/cli@latest whoami

Claude Code plugin.

Brings the same skill plus the command /time-momentum:setup. Run both commands inside Claude Code.

  1. Add the marketplace

    /plugin marketplace add https://time-momentum.com/claude/marketplace.json
  2. Install the plugin

    /plugin install time-momentum@time-momentum

Command groups.

Every API operation has exactly one command. tm schema <group> --json lists flags, types and examples.

GroupWhat it covers
tm timersStart, pause, stop and log time entries
tm customers, tm projects, tm ordersClients, projects with budgets, orders and their billing type
tm invoices, tm paymentsBuild, issue and export invoices as e-invoices, record and split payments
tm statementsTimesheets as invoice attachment or on their own
tm ai-sessions, tm progressAgent sessions with runtime and usage, project progress
tm revenue, tm dashboard, tm exportProfitability, key figures and raw data as CSV or Excel
tm collectThe agent collector: connect tools, sync, subscription limits
tm apiAny endpoint directly, signed in like every other command

REST API.

tm is a client of the REST API at https://app.time-momentum.com/api/v1. Business logic such as totals, taxes and invoice numbers lives on the server, so scripts and agents get the same results as the web app.

An API key (tmk_…) works directly as a bearer token on the read routes, the timers and the collector ingest. A key with write access is exchanged for a token valid for one hour with POST /api/v1/auth/token. tm does this for you and renews it automatically.

Keys are created only in the web app and can be revoked there at any time. The server refuses to create keys or delete the account through agent access.

Every operation with parameters, request and response schemas: OpenAPI specification (OpenAPI 3.1, descriptions in German).

For agents and scripts.

  • llms.txt: pages and setup guides of this website in one file
  • agents/collector.md and agents/cli.md: setup guides for AI agents
  • tm --json: compact overview of all command groups, conventions and exit codes
  • openapi.yaml: the REST API as OpenAPI 3.1
  • tm schema --json: the full command catalog with flags, types and endpoints
  • tm skill: the usage guide for agents, matching the installed version

Connect your tools today.

  • 30 days of Pro
  • No credit card
  • Then continue on Free

Frequently asked questions

Do I need Pro?

The agent collector, the timers and the project list work in Free. Write access for the command line and your agent, invoices and the other functions are part of Pro, which every new account tests for 30 days.

Why does my agent not ask for the API key?

So the key never ends up in the conversation. The agent installs tm and asks you to run tm collect init or tm auth login in a terminal of your own, where the key is entered with hidden input.

Is there an MCP server?

No. Your agent already works in a shell, and tm loads only the part of the command catalog it needs. More in the guide MCP vs. CLI for AI agents.