---
title: Install
description: Install ClawMetry with pip, the one-line installer, Homebrew-style scripts, Docker, or the macOS desktop app — plus what each method actually sets up.
keywords: install ClawMetry, pip install clawmetry, ClawMetry docker, ClawMetry windows install
eyebrow: Get started
---

# Install

ClawMetry is a Python package with four dependencies. The install puts a
`clawmetry` command on your path and, on first run, sets up a background sync
daemon.

## Requirements

| | |
|---|---|
| **Python** | 3.9 or newer (CI covers 3.9 and 3.11 on Linux, macOS and Windows) |
| **Disk** | The DuckDB store grows with your agent activity; a few hundred MB is typical. [Retention](/docs/data/retention/) bounds it. |
| **Network** | None required. Cloud sync is opt-in. |
| **Agents** | Anything from the [runtime list](/docs/runtimes/overview/). ClawMetry finds them itself. |

## pip

The direct route, and the one every other method wraps.

```bash
pip install clawmetry
clawmetry onboard
```

`clawmetry onboard` is the first-run wizard: it detects your runtimes, offers
local-only or cloud mode, starts the sync daemon, and opens the dashboard. See
[onboard / setup](/docs/cli/onboard/) for every flag, including the
non-interactive forms.

To stay entirely local and never be asked about an account:

```bash
clawmetry onboard --local
```

## One-line installer

Detects the OS, installs into a sensible interpreter, and starts ClawMetry.

```bash title="macOS, Linux, WSL"
curl -fsSL https://clawmetry.com/install.sh | bash
```

```powershell title="Windows PowerShell"
iwr -useb https://clawmetry.com/install.ps1 | iex
```

```bat title="Windows CMD"
curl -fsSL https://clawmetry.com/install.cmd -o install.cmd && install.cmd && del install.cmd
```

:::warning Two Python installs
The single most common install problem is a machine with more than one Python:
`pip install clawmetry` lands in one interpreter and `clawmetry` resolves to a
shim from another. If the version you get from `clawmetry --version` is not the
version you just installed, run `python3 -m pip install --upgrade clawmetry` and
check `which -a clawmetry`.
:::

## Docker

The container has no access to your host's agent directories unless you mount
them, so mount whichever runtimes you want observed, read-only.

```bash
docker build -t clawmetry .
docker run -p 8900:8900 \
  -v ~/.openclaw:/root/.openclaw:ro \
  -v ~/.claude:/root/.claude:ro \
  clawmetry
```

See [Docker](/docs/config/docker/) for volume layout, persisting the store
across restarts, and the environment variables that matter in a container.

## macOS desktop app

A native menubar shell that pip-installs and supervises the same package. It
runs alongside the CLI rather than replacing it — the daemon, store and
dashboard are identical. Download it from the
[homepage](https://clawmetry.com/#get-started).

## Ask your agent to do it

The setup is interactive, so run it in a long-lived terminal session rather
than a one-shot command:

```text
Help me install ClawMetry in a long-running PTY:
pip install clawmetry && clawmetry setup && set up clawmetry sync via supervisord
```

## Verify the install

```bash
clawmetry --version          # the package version
clawmetry status             # daemon state, store size, last ingest
clawmetry runtimes           # what was detected on this machine
clawmetry diagnose           # entitlement + resolver inputs, if something looks locked
```

Then open <http://localhost:8900>.

If `clawmetry status` says the daemon is not running, start it with
`clawmetry sync`. If the dashboard loads but every tab is empty, the daemon is
the thing to look at first — the dashboard reads through it and shows nothing on
its own. [Troubleshooting](/docs/ops/troubleshooting/) walks the whole chain.

## Keeping it current

```bash
clawmetry update             # or: pip install --upgrade clawmetry
```

The daemon also checks PyPI on a cadence and can update itself. That behaviour,
and how to turn it off, is documented in [Upgrades](/docs/ops/upgrades/).