The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill generates and renders HTML templates that may incorporate untrusted external data (e.g., social media content, user inputs) during the design process.
Ingestion points: The 'Generate HTML' workflow step (Step 1) accepts arbitrary content to be styled and rendered within an HTML/CSS template.
Boundary markers: No specific boundary markers or instructions to ignore embedded scripts/instructions within the HTML are provided.
Capability inventory: The skill uses Playwright for browser operations (navigate, screenshot) and shell commands for server management.
Sanitization: No sanitization or escaping of the input data used to construct the HTML template is mentioned.
[COMMAND_EXECUTION]: The skill executes shell commands to manage a local HTTP server used to serve the generated HTML files to Playwright.
Evidence: The core workflow includes shell commands: python -m http.server 8765 --directory "OUTPUT_DIR" & to start the server and pkill -f "http.server 8765" to stop it.

image-creator