Skip to main content
AI Choice Engine
← All setup guides

Meta · Setup guide

How to set up Muse Code

Install Muse Code on a documented platform, confirm the credential, model and permission profile, then try one bounded task in a disposable repository. Terminal, CI and session-SDK surfaces share the agent, but workflows depend on the build and rollout.

Muse Code

terminal · checked October 2, 2026

~12 minIntermediate
Start installOpen Muse CodeCompare itBuyer guide

The “Open Muse Code” link goes to the vendor site. It is not an affiliate-tracked link on this page — no commission is claimed for that click. Affiliate disclosure.

Critical: finish install before closing the window

Keep the installer terminal open until you see a success message. If you used an admin window, do not close it until verification works. Then open a new Terminal or PowerShell window and run the verify step — PATH updates rarely apply to shells that were already open, especially on Windows.

Before you begin

  • macOS, Linux, or Windows (the Muse Code docs list all three; the product page lists macOS and Windows).
  • A Meta account with eligible Muse Code access, or a Meta Model API key and billing. Meta Managed Accounts require API-key authentication. Use a disposable repository for the trial.
  • A shell, Git, and a clean branch for the first task.

Authentication

Meta account sign-in inside the CLI

Account: Meta account with Muse Code / Model API access

Browser sign-in or an API key is documented; Meta Managed Accounts require a key. META_API_KEY overrides a stored key, which overrides browser sign-in. A subscription covers its automatically linked CLI credential; additional API keys bill pay-as-you-go. Verify the selected model: current docs default to muse-spark-1.2, even though 1.3 is available.

Install or open it

1. Get the harness ready

These steps are written for beginners. On Mac, Terminal is the black/white text window where you type commands. On Windows, use PowerShell or Terminal. Use the OS tabs to hide steps that are not for your computer, then copy one command at a time.

  1. 01

    Open Terminal on a Mac

    Mac

    Click the Spotlight search icon (magnifying glass) at the top-right of the menu bar, or press Command (⌘) + Space. Type Terminal and press Return. A text window opens — that is where you paste the install commands below. Leave it open.

  2. 02

    Open Terminal or PowerShell on Windows

    Windows

    Press the Windows key, type Terminal or PowerShell, then press Enter. Use that window for the commands below. If Windows asks for permission to run a script or install software, read the prompt before approving.

  3. 03

    How to paste a command

    Copy one command at a time from this page. Click inside the terminal window, paste (Mac: ⌘+V, Windows: Ctrl+V or right-click → Paste), then press Return / Enter. Wait until the prompt comes back before running the next command. You do not need to type the leading $ if you see one in other docs.

  4. 04

    Do not close the installer until verified

    Keep the installer terminal (or admin PowerShell) open until the install finishes without errors. Closing early can leave a half-written binary or an incomplete PATH update — especially on Windows.

  5. 05

    Open a new terminal after installers

    PATH updates often only apply to new shells — especially on Windows. Do not close the install session until the installer finishes, then open a fresh Terminal/PowerShell before running --version checks.

  6. 06

    Inspect before you pipe an installer

    Commands like curl … | bash download and run a script immediately. Prefer the official package manager when offered. If you use a curl installer, you can download the script first, open it in a text editor, then run it — never paste installers from random blogs.

  7. 07

    Install on Mac / Linux

    Paste Meta’s official installer into Terminal. Prefer Meta’s current product docs if the installer URL changes.

    bash
    curl -fsSL https://dev.meta.ai/install.sh | bash

    You should see: The installer finishes without an error. Open a new terminal before the verify step so PATH updates apply.

    If this fails
    • The install command errors, or the next verify step says “command not found”. Re-run the official install command with network access, then open a brand-new terminal window before verifying.
    Muse Code product page
  8. 08

    Install on Windows (PowerShell)

    Windows

    Paste Meta’s official PowerShell installer, then open a new terminal window before verifying.

    powershell
    irm https://dev.meta.ai/install.ps1 | iex

    You should see: The installer finishes without an error. Open a new terminal before the verify step so PATH updates apply.

    If this fails
    • The install command errors, or the next verify step says “command not found”. Re-run the official install command with network access, then open a brand-new terminal window before verifying.
    Muse Code docs
  9. 09

    Check the command

    Open a new terminal window, then confirm the binary before entering a real repository.

    bash
    muse --version

    You should see: A version string. Open a new terminal if the command is not found, then confirm your Meta account has Muse Code access.

    If this fails
    • The installer completes but muse is unavailable. Open a new shell, check the installer directory on PATH, and compare the installed command with Meta’s current instructions.
    Muse Code docs
  10. 10

    Authenticate before any repo init

    Run `muse` in a disposable folder first. Complete Meta sign-in in the TUI (`/login`), or set a key with `muse auth set` / `META_API_KEY`. Do not run `muse init` on a production tree until auth succeeds.

    bash
    muse

    You should see: The Muse TUI opens and prompts for Meta login or an API key — not an unauthenticated init.

    If this fails
    • Auth fails, the key is rejected, or the provider list is empty. Confirm the key/account is active, that you pasted the whole value, and that you are using the shell syntax for your OS (export vs $env:).
    Muse Code authentication

Repository or project

2. Start in the right place

Keep the first repository small, clean, and easy to roll back while you learn the harness’s permissions and context model.

  1. 01

    Use a disposable branch

    Choose a small repository or branch with tests and no production credentials.

    bash
    git switch -c muse-code-pilot

    You should see: The prompt shows you are inside the intended folder (check with `pwd`).

    If this fails
    • “No such file or directory” or you are still in your home folder. Create or clone the practice folder first, then `cd` into it before starting the tool.
  2. 02

    Initialise AGENTS.md only after auth

    From the repository root, use muse init --dry-run to preview AGENTS.md. Ordinary muse init refuses to overwrite an existing file; --force overwrites it completely. Preserve existing instructions and review any new file before trusting the workspace.

    bash
    muse init --dry-run

    You should see: The prompt shows you are inside the intended folder (check with `pwd`).

    If this fails
    • “No such file or directory” or you are still in your home folder. Create or clone the practice folder first, then `cd` into it before starting the tool.
    Muse Code configuration

First run

Review the agent’s event trail

Use this task to test the workflow before handing over a larger repository or production responsibility.

  1. 01Choose Ask me or Read-only explicitly for the first exploration; Auto-review is the default and can approve eligible requests automatically.
  2. 02Ask for a repository explanation and one small proposed change. Normal workspace file writes and recognized safe shell stages can pass without asking, so a prose request alone does not create a read-only boundary.
  3. 03Review the event log, diff, and test result before trying a second task.

Why this is safe: A bounded pilot exercises event visibility, permissions and the final diff. Retained session logging can be disabled, so confirm that the event trail you plan to inspect is enabled.

Stop and review if

  • The agent cannot show which worker or command produced a change.
  • The task starts background work outside the repository boundary.
  • The model or contributor data setting is unclear for the repository.

The operating boundary

Privacy, data, and cost

The setup is incomplete until you know what the harness can see, what account pays for it, and who approves a risky action.

Privacy

Data use follows the selected model, including subscription use. Standard Services exclude model training; Discounted/Contributor Services permit it, and Meta's terms forbid submitting confidential, sensitive or personal information to them. Standard no-training does not promise zero retention; service, safety and legal processing remain.

Training on code: tier dependent

Confirm Meta's current terms, model designation and geographic eligibility. The September 18 API terms describe a limited preview period; an unlabeled product page does not establish universal availability or a completed GA transition.

Sensitive repos: Start with a sanitized repository, default to standard (non-contributor) rates for confidential code, and do not connect production secrets to a new harness during a pilot.

Pricing

October 2 product pricing: Everyday $5/month, High $15 (5x) and Power $50 (20x), with regional benefits varying. The CLI-linked subscription credential is separate from additional pay-as-you-go keys. Spark 1.3/1.2 Standard API rates are $1.25 input / $0.15 cached / $4.25 output per MTok; Contributor is $0.10 / $0.002 / $0.20.

Model: hybrid

Watch out: API quotas are per team: Standard 3,000 RPM / 4M TPM; Contributor 100 RPM / 3M TPM. Web-search grounding adds $2.50 per 1,000 queries. Subscription allowances expire without rollover and additional keys remain pay-as-you-go; choosing a Contributor model changes data use without increasing the subscription allowance.

Estimate a seat stack

When setup goes wrong

  • The installer completes but muse is unavailable.

    Open a new shell, check the installer directory on PATH, and compare the installed command with Meta’s current instructions.

  • Muse Code access is unavailable after sign-in.

    Confirm the account, region, and subscription or API billing; do not treat a successful install as proof of product access.

  • Background work is difficult to review.

    Stop owned background work and inspect the retained event log and diff. Confirm that workflows are enabled for the build: the documented current Apple silicon macOS package lacks the workflow engine. Windows lacks session messaging and voice; those missing features do not imply failed authentication.

Next steps

After the first successful run

Keep the pilot small, then compare surfaces and budget before a team rollout.

Compare harnessesBuyer guideOpen-source agents hubAI stack cost calculatorGPU inference (rent)Local GPU sizingBuyer guidesLocal open-source AI guide

Setup details change quickly. The page was checked October 2, 2026; use the official source below before installing in a managed or production environment.

Muse Code docsRead the launch guideCompare all harnesses