Skip to main content
AI Choice Engine
← All setup guides

Amp · Setup guide

How to set up Amp

Install Amp beside the editor you already use, sign in at ampcode.com, then practice only on a disposable clone — by default Amp runs tools with no approval prompt. This guide covers the local CLI, web sign-in, optional SDK tokens, Attach Anything context, and the billing boundary between subscriptions, credits, and remote Orbs.

Amp

terminal · checked October 2, 2026

~10 minBeginner
Start installOpen AmpCompare itBuyer guide

The “Open Amp” 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 or Linux; CLI docs name Windows through WSL, while a first-party native PowerShell installer also exists. Confirm the supported installed-client path before rollout.
  • A terminal and a disposable git clone you can safely delete (not a production working tree).
  • An Amp account at ampcode.com and, if needed, a paid subscription or credits.
  • Secrets removed from the practice clone; Private thread visibility if you are in a shared workspace.

Authentication

Amp account at ampcode.com; amp login adds a CLI account

Account: Amp account; a subscription or credits may be needed for agent usage

Create or sign in at ampcode.com and run amp from a practice repository. Current CLI docs document amp login for adding an ordinary CLI account, amp account list/switch for account selection, and AMP_API_KEY for environment-based authentication. The environment key overrides saved accounts. Account switches affect future processes, not already-running sessions; keep tokens out of repositories and shell history.

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 Amp on macOS, Linux, or WSL

    Paste the official installer into Terminal. It installs the Amp executable for the current platform; when it finishes, open a fresh terminal before checking the command.

    bash
    curl -fsSL https://ampcode.com/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.
    Amp Owner’s Manual
  8. 08

    Install Amp on Windows

    Windows

    This first-party script downloads a native Windows binary. Current CLI prose separately says Windows via WSL; native support scope is unresolved. Prefer the documented WSL path unless the current installer/client explicitly supports your environment, then reopen the shell for PATH.

    powershell
    powershell -c "irm https://ampcode.com/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.
    Amp Owner’s Manual
  9. 09

    Optional: install with Homebrew on a Mac

    Mac

    If you already use Homebrew, this official tap is an alternative to the direct installer. Do not install an unrelated package that merely happens to use the name amp.

    bash
    brew install ampcode/tap/ampcode

    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.
    Amp Owner’s Manual
  10. 10

    Verify the CLI

    Amp uses its help screen as the simple install check. If the shell cannot find amp, open a new terminal and check the installer’s PATH message.

    bash
    amp --help

    You should see: Amp prints its usage and available commands. If the command is missing, open a new terminal and inspect PATH.

    If this fails
    • amp: command not found after the installer finishes. Open a new terminal first. If it still fails, rerun the official installer and add the printed install directory to PATH; avoid installing an unrelated npm package named amp.
    Amp Owner’s Manual

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

    Enter a disposable practice clone

    Use a throwaway clone, not your production working tree. On a Mac, type cd followed by a space and drag the folder from Finder into Terminal.

    bash
    cd path/to/your-disposable-clone

    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

    Start Amp (after web sign-in)

    Sign in at ampcode.com in the browser first if you have not already. Then run Amp from the disposable clone root. Threads in workspace-owned projects default shared with workspace members; personal/no-project threads default private, subject to admin controls. Review audience before the first prompt.

    bash
    amp

    You should see: The tool starts in this folder and waits for your first prompt (or shows a short first-run / sign-in screen).

    If this fails
    • Nothing happens, or you land in the wrong folder. Confirm you are inside the practice repo (`pwd`), that you signed in, and that you opened a new terminal after install.
    Amp Owner’s Manual
  3. 03

    Optional: SDK or headless token login

    Use amp login when adding a CLI account; it is not restricted to SDK use. For headless automation, provide AMP_API_KEY through the environment. It takes precedence over saved accounts, so check which identity future CLI processes will use.

    bash
    amp login

    You should see: The shell accepts the command with no error so the next step can use the credentials or provider.

    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:).
    Amp SDK authentication

First run

Disposable clone, explicit no-edit orientation

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

  1. 01Work only in a disposable clone. Review visibility before the first prompt: workspace-owned projects default shared, while personal/no-project threads default private, subject to admin controls.
  2. 02Know Amp’s default: it does not ask for approval before running tools. Use a reviewed custom policy plugin if your organisation requires gated tool use; the current tools docs do not promise a legacy permission mode.
  3. 03Ask: explain the entry points, test command, and likely deployment boundary. Explicitly say: do not edit files, do not run shell commands, and do not enable MCP or Orbs.
  4. 04If Amp still runs a tool or edits a file, stop the thread, inspect git status, and discard the clone if needed.

Why this is safe: Safety comes from the disposable clone and an explicit no-edit / no-command prompt — not from an approval gate Amp does not use by default.

Stop and review if

  • You are on a production branch or any clone you are not willing to delete.
  • The thread is Workspace-shared or Unlisted while the repository is private.
  • Amp edits files, runs shell commands, enables MCP, or starts an Orb despite the no-edit prompt.
  • The agent asks for production credentials.

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

Amp stores selected code/tool output in threads; Orbs clone the project and retain paused filesystem/memory until deletion. Workspace-owned project threads default shared; personal/no-project threads private, subject to admin controls. Managed inference MDR has safety/legal/model exceptions and excludes Amp thread storage. Training requires explicit opt-in and cannot be enabled for Enterprise. BYOK and linked consumer subscriptions follow separate provider terms.

Training on code: tier dependent

Workspace, enterprise, model-provider, and remote Orb terms can change the data path; confirm current Amp terms before regulated code.

Sensitive repos: Set visibility to Private (or leave a shared workspace) before the first practice thread, remove secrets, review .amp settings and MCP servers, and stay on a sanitized clone until your organisation approves retention and remote execution.

Pricing

Hobby client access is free with BYOK/linked subscriptions and paid Orbs; Megawatt is $20/month with 45,000 weighted Orb minutes. Teams adds no extra charge; Enterprise is custom, and Gigawatt price is unpublished here. Orbs bill by minute at $0.08–$2.13/hour, with the largest size restricted, and pause only after both 5 minutes without agent work and 20 minutes without interaction. Included usage does not roll over; purchased credits expire after 12 months. Non-enterprise paid-credit inference has no API markup, but non-model tools can also consume credits. Education is $10/month; no extra credit grant is confirmed.

Model: hybrid

Watch out: Check the active executor, workspace, model pins, Orb size and visibility. ChatGPT linking alone does not reroute every mode role; the ChatGPT Only preset pins low/medium/high roles to eligible OpenAI models, including Astra high/oracles. Ultra keeps Amp routing. BYOK Fable needs Anthropic retained-data workspace access. Opening a thread/portal can wake an Orb and restart its interaction timer; paused storage is retained.

Estimate a seat stack

When setup goes wrong

  • amp: command not found after the installer finishes.

    Open a new terminal first. If it still fails, rerun the official installer and add the printed install directory to PATH; avoid installing an unrelated npm package named amp.

  • amp cannot find an account or asks you to sign in.

    Complete sign-in at ampcode.com in the browser, then rerun amp from the practice clone. Use amp login to add a CLI account; AMP_API_KEY is an environment override for headless use and takes precedence over saved accounts. Keep tokens out of shell history and git.

  • The thread exposes code more widely than intended.

    Stop the run, change visibility to Private (or leave the shared workspace), remove untrusted workspace MCP entries, and start again from a sanitized clone. Archived or shared threads can remain accessible by URL.

  • Amp edits files or runs tools during a “read-only” prompt.

    Expect that: Amp does not ask for approval by default. Stop the thread, inspect git status, discard the disposable clone if needed, and configure a reviewed policy plugin before retrying on another throwaway clone.

  • Amp asks for payment or an Orb before the first task.

    Confirm whether the thread is using local execution, a subscription, workspace credits, or an Orb. Keep the pilot local and short until billing ownership is clear.

Next steps

After the first successful run

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

Compare harnessesBuyer guideAmp vs Claude Code (first week)AI 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.

Amp Owner’s ManualCompare all harnesses