Check your app on your own machine, free
Stunt Double Studio is a free local app. Your coding agent runs a checklist in Chrome on your machine, against localhost, staging or production, and tells you which checks fail, with the evidence.
What Studio gives you
Studio verdicts are judged by your own agent. A cloud run adds an independent judge.
A real browser run for every check
Each check runs in Chrome on your machine, against localhost, staging or production. A pass must cite text or an element on the page, and the viewer keeps a screenshot of every step.
Runs kept in your project
Every run is a set of files in .stuntdouble/, next to your code. Commit them, diff them between branches, or hand them to a teammate.
A viewer to watch and stop a run
The local viewer shows each verdict, its screenshots and the activity log as the run goes. Stop the run from there when you have seen enough.
Free, on your own subscription
Studio is free. Local runs use the Claude plan you already have, or your API key if you set one. Stunt Double makes no model calls for a local run, and you need no Stunt Double account.
Studio needs Node 22.18 or later and Google Chrome (or Playwright's Chromium).
On production, the actor does what a person would, so use test accounts. Screenshots in .stuntdouble/runs are committed with everything else.
Install in two commands
In a terminal, add the marketplace, then install the plugin. Then start a new Claude Code session, since plugins load when a session starts.
Using another coding agent? Add the Studio MCP server over stdio with npx -y stuntdouble mcp. The stuntdouble package is on npm under Apache-2.0. The plugin also adds the hosted server for cloud runs. You can leave it signed out, since local runs don't use it.
Or ask your agent to do it all
Paste this into Claude Code. Your agent installs the plugin, then you start a new session and run setup.
Install the Stunt Double plugin for Claude Code (claude plugin marketplace add stunt-double/stuntdouble-mcp, then claude plugin install stuntdouble@stuntdouble). When it is in, tell me to start a new Claude Code session and run /stuntdouble setup for my sign-up flow on http://localhost:3000, as a first-time visitor.Commands
Run these from Claude Code once the plugin is in.
| Command | What it does |
|---|---|
/stuntdouble setup | Sets up Studio in this project. |
/stuntdouble run <checklist> | Runs a checklist in Chrome on your machine. |
/stuntdouble open | Opens the local viewer: verdicts, screenshots and the activity log. |
/stuntdouble doctor | Checks Node, Chrome, the viewer port and how local runs are billed. |
/stuntdouble cloud <checklist> | Moves a checklist to a Stunt Double workspace and starts a cloud run. Signs you in the first time. |
Run the same checklist on every deploy
Studio runs one checklist at a time, while your agent session is open. Cloud runs take the same checklist further: /stuntdouble cloud moves it to a workspace with the same actor and checks. A localhost URL needs a public or preview URL for the cloud.
- Parallel, unattended and scheduled runs
- Runs on every deploy, pull request or Figma change
- An independent judge on every verdict
- Interviews, design reviews and feedback triage
- Index scoring
- Team sharing, with Slack, Linear and GitHub
The Free cloud plan includes 60 compute minutes a month and 2 actors, with no card. Free cloud runs reach public URLs only, and may be used to improve actors.