The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use shell commands to interact with the local development environment and execute tests.
The agent is guided to use netstat, lsof, or curl to identify the active Vite development port on localhost before initiating browser-based interactions.
The test suite is executed via npm run test:e2e, which is a standard mechanism for running Playwright tests in a Node.js project.
[PROMPT_INJECTION]: The skill establishes a surface for processing untrusted qualitative data through its visual analysis helpers, characteristic of indirect prompt injection risks.
Ingestion points: Design specifications (GDD) are accepted as plain text in functions like checkGDDCompliance to be evaluated against screenshots.
Boundary markers: The provided code templates do not implement specific delimiters or safety instructions to isolate these external design specifications from the agent's core operational instructions.
Capability inventory: The agent possesses browser automation capabilities through Playwright (page.evaluate) and shell command execution.
Sanitization: No input validation or filtering of the GDD specification text is described before it is passed to the analysis helpers.

qa-visual-testing