The Agent Skills Directory

[COMMAND_EXECUTION]: Executes shell commands via npx playwright to generate screenshots of local routes (e.g., http://localhost:{port}/{route}). The command utilizes variable interpolation for port numbers and route paths derived from the project configuration or user input.
[EXTERNAL_DOWNLOADS]: Recommends installing the playwright package via standard Node.js package managers (npm) and utilizes npx for on-demand execution. These are established workflows for well-known developer tools.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from modified source files and visual screenshot content to drive the verification loop.
Ingestion points: Reads modified view-layer files (.tsx, .jsx, .vue, .svelte, .html, .css) and processes vision data from generated screenshots.
Boundary markers: No explicit delimiters or instructions are provided to the agent to ignore potentially malicious content within the analyzed files or screenshots.
Capability inventory: Includes shell command execution (npx playwright) and the ability to modify source code files during the "fix" phase.
Sanitization: No specific sanitization or validation logic is defined for the route names or code fixes generated from external inputs.

live-preview