Skip to content

Docs for the AI testing agent, MCP server and visual testing

Your first AI-written test in 5 minutes

Point the agent at your app's URL, read the verdict, and keep the tests as Playwright code in your own repository. Everything from sign-in to CI.

Start the 5-minute quickstart Connect your coding agent (MCP)
Add Wopee.io to Claude Code 
# in your repository, with WOPEE_API_KEY and WOPEE_PROJECT_UUID exported
claude mcp add --scope project wopee -e 'WOPEE_API_KEY=${WOPEE_API_KEY}' -e 'WOPEE_PROJECT_UUID=${WOPEE_PROJECT_UUID}' -- npx -y wopee-mcp
# then start claude and ask
"Use Wopee.io to analyze my app and generate tests. Focus on the checkout flow."

Key and project UUID: where to find them · Other agents: Copilot, Cursor, Codex and more

What do you want to do?

Pick a path. Every card leads to the page that finishes the job.

Four steps from URL to verdict

What the quickstart walks you through.

  1. Sign inWith GitHub, GitLab or Google at cmd.wopee.io.
  2. Create a projectPaste your app's URL, or pick a demo app.
  3. Watch the agent exploreIt clicks through your app and writes user stories and test scenarios.
  4. Read the verdictEvery run has a report: the verdict, each step and a screenshot.
Play: Wopee.io in 2 minutes

See it in 2 minutes

An AI agent tests a web app, from the URL to the verdict.

More walkthroughs: quick start in 6 minutes, Claude Code with the MCP server, your tests in your repo.

Bring your own framework

Visual checks inside the tests you already have.

Playwright Cypress WebdriverIO Robot Framework Any stack: GraphQL API

Limited support also exists for mobile apps, PDFs, desktop apps and images your application generates. Contact us for details.