Verified Agent Skill record

playwright-cli

Automate browser interactions, test web pages and work with Playwright tests. Provides commands for navigation, element interaction, snapshots, screenshots, PDF export, tab management, storage state, network mocking, and DevTools access.

Tasks

Fill and submit a login form using snapshot refs: run playwright-cli open, snapshot to get element refs, fill the email and password fields, click submit, then snapshot again to verify the result
Capture a full-page screenshot and PDF of a web page using playwright-cli screenshot --filename and playwright-cli pdf --filename

Inputs

URLs to navigate to via playwright-cli goto or playwright-cli open
Snapshot refs (e.g. e15) or CSS selectors or Playwright locators to target elements for click, fill, type, and other interactions
File paths for screenshots (--filename=page.png), PDFs (--filename=page.pdf), storage states (auth.json), and trace/video output
Global flags such as --raw for piping output, --json for structured JSON output, and -s=<session> for named browser sessions
Browser selection via --browser=chrome, --browser=firefox, --browser=webkit, or --browser=msedge

Outputs

Page snapshots in YAML format with page URL, title, and element refs
Screenshot files in PNG format
PDF files
Console logs and network request logs from DevTools commands
Raw or JSON-formatted command results when using --raw or --json flags

Limitations and checks

No security conclusions may be drawn from the source document; the skill file describes functionality but does not document access controls, sandboxing, or threat models
The source document does not specify which operating systems or Node.js versions are supported
References to additional guides (playwright-tests.md, request-mocking.md, etc.) are listed but their contents are not included in the source document
The source document does not document how playwright-cli is authenticated or authorized to control browsers
Verify playwright-cli is available before starting by running npx --no-install playwright-cli --version or playwright-cli --version
After every navigation or interaction command, read the returned snapshot to confirm the page state and obtain refs for subsequent actions
When using --raw to pipe output, verify the consuming command (e.g. jq or diff) is available and that the raw output format matches expectations
If using named sessions with -s=<session>, ensure you explicitly close each session with playwright-cli -s=<session> close to avoid orphaned browser processes