How to Use Muse Code

Muse Code is Meta's CLI coding agent released 2026-08-05. This guide walks through install, browser authentication, first task, and using the event log.

The install and auth flow is deliberately minimal. You should be running a real task within five minutes of the curl command.

Every command below is quoted from Meta's launch materials. Anything not published (headless auth, MCP wiring) is called out.

Install Muse Code

Install with a single line: curl -fsSL https://dev.meta.ai/install.sh | bash. Meta ships this as the primary install path.

The script drops the muse binary into your PATH. Verify with muse --version.

For teams that want to review the install script before running it, download it first (curl -fsSL https://dev.meta.ai/install.sh -o install.sh), audit, then run.

• curl -fsSL https://dev.meta.ai/install.sh | bash
• Verify: muse --version
• Optional: download + audit script first
• Binary lands in $PATH

Browser Authentication

First run of muse triggers a browser auth flow against your Meta developer account. Sign in and grant the CLI access.

For CI or headless environments, a documented headless auth flow is not available at launch. Verify at https://developer.meta.com/ai/products/muse-code/.

Once authenticated, credentials cache locally. Subsequent runs skip the browser step.

• First run: browser auth against Meta developer account
• Headless / CI flow: not documented
• Credentials cache locally after first auth
• No re-auth per session

Your First Task

cd into a repo, then run muse. The agent reads the local context and starts a task loop.

Give it a small verifiable task: "add a missing test for the parseDate function and make sure it passes." You will see a worker draft the test, a reviewer critique it, and the test run.

The whole round trip usually takes a couple of minutes. This is the fastest way to feel the multi-agent workflow in practice.

• cd into a repo, run muse
• Start with one small verifiable task
• Worker drafts → reviewer critiques → test runs
• A few minutes to full round-trip

Using the Event Log

Every session records a full event log — file edits, tool calls, and decisions. Run muse replay to walk it step by step.

Export the full trace when you need to share a session with a teammate or attach it to a compliance review.

In our engagement with client engineering teams we treat the event log as a first-class code-review artifact. It gives an outsider enough context to review the agent's reasoning without re-running the session.

• muse replay = step-by-step walk
• Export the trace for handoff / audit
• First-class code-review artifact
• Shareable with teammates

How to use Muse Code

You do not host Muse Code yourself — you use it through a tool, so "getting started" really means choosing the right one.

The fastest way to put Muse Code to work day to day is inside an AI IDE, and Cursor is the most popular — it supports it directly, so you can be working in minutes. Prefer a different editor? Windsurf, Zed, and GitHub Copilot drive these models too.