The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructions create a surface where malicious content from external sources could influence agent behavior.
Ingestion points: The agent is directed to explore live demo websites using agent-browser and download WordPress theme files (CSS, PHP, JSON) via curl as reference material.
Boundary markers: The skill lacks explicit delimiters or instructions to ignore embedded commands within the source material it processes.
Capability inventory: The agent has access to shell execution (pnpm dev, pnpm build), network operations (curl), and browser automation (agent-browser eval).
Sanitization: No sanitization or validation logic is defined for the content extracted from third-party themes before it is used to guide the porting process.
[EXTERNAL_DOWNLOADS]: The skill performs network downloads of external archives and assets.
Evidence: In phases/1-discovery.md, the agent is instructed to fetch theme ZIP files from downloads.wordpress.org and download images from arbitrary demo site URLs provided by the user or discovered during the task.
[COMMAND_EXECUTION]: The skill uses shell commands to manage files and execute development scripts.
Evidence: The agent uses unzip to extract theme source code and pnpm to run a local development server and build the project (pnpm dev, pnpm run build).
[DYNAMIC_EXECUTION]: The skill executes dynamic JavaScript within a browser context to extract design tokens.
Evidence: In phases/2-design.md, the agent uses agent-browser eval to execute a script that extracts computed styles (fonts, colors, and layout metrics) from the target WordPress demo site.

wordpress-theme-to-emdash