Tests that read like user stories.
Each step has an intent that describes the user goal. Locators are cached for speed, but when they break, the AI re-derives the action from the intent — no manual fix needed.
goal: Verify user can create a new project
base_url: https://app.example.com
statements:
- URL: /projects
- intent: Click the "New Project" button
- intent: Enter "My Test Project" in the project name field
- intent: Click "Create"
action: click
locator: "getByRole('button', { name: 'Create' })"
- VERIFY: Project page shows title 'My Test Project'Core Features
Intent-driven execution
Tests describe what users do, not how the feature is built. Implementation changes don't break tests — only behavior changes do.
Self-healing by design
Tests are resilient to non-functional changes. Real behavior changes are surfaced as failures.
Fits your existing workflow
Runs alongside your existing Playwright tests and configuration. Add YAML tests incrementally — no migration, no new tooling to learn.
Part of the
Shiplight Plugins
Browser MCP server and YAML tests work together. Your agent uses the MCP server to automate development, skills guide it to write YAML tests, and the Shiplight runtime runs them.
Browser MCP Server
Eyes and hands for your agent. Automates browser interactions during development.
Skills
Guide your agent to write thorough, self-healing YAML tests from browser MCP interactions.
Shiplight Runtime
Executes YAML tests, provides feedback to your agent, generates comprehensive reports, and supports local debugging.