The Agent Skills Directory

[PROMPT_INJECTION]: Indirect prompt injection vulnerability surface.
Ingestion points: The skill ingests untrusted data from external URLs (Stage 1) and the 21st.dev component registry (Stage 2) to generate code.
Boundary markers: No explicit instructions are provided to the agent to treat external content as data only or to ignore embedded instructions within the fetched HTML or components.
Capability inventory: The skill has Write, Edit, and Bash permissions, which are used to modify project source code based on the external input.
Sanitization: There is no evidence of sanitization or validation of external content before it is integrated into the codebase.
[EXTERNAL_DOWNLOADS]: Fetches design context and component definitions from well-known services.
Evidence: The skill integrates with Google's Stitch MCP (stitch.withgoogle.com) for design extraction and the 21st.dev registry for component matching.
[COMMAND_EXECUTION]: Performs automated file system and verification tasks.
Evidence: Uses Bash, Glob, and Grep for project context discovery and run-story-tests for component verification.

design-to-code