The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data to automate file-system operations.
Ingestion points: The skill reads from ./docs/PRD.md and ./docs/APP_FLOW.md to detect project types and extract screen inventories.
Boundary markers: Absent. There are no delimiters or instructions provided to the agent to disregard potentially malicious instructions within the documentation files.
Capability inventory: The skill possesses the capability to write numerous files (React components and page shells) to the project directory.
Sanitization: Absent. The skill does not validate or sanitize screen names or indicators found in the documentation before using them to generate file paths, which creates a potential surface for path traversal if the documentation is maliciously crafted.

vibe-coding-design-templates