The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-supplied requirement descriptions that influence the generated content of local files. \n- Ingestion points: User requirement descriptions provided during the initial analysis and clarification rounds in Step 1 and Step 3 of SKILL.md. \n- Boundary markers: No specific delimiters or boundary markers are instructed to separate user input from the skill logic within the instructions. \n- Capability inventory: The skill uses a Write tool to save documentation in the local ./docs/prds/ directory. \n- Sanitization: The instructions do not define explicit sanitization or validation for user input text. \n- [SAFE]: The skill is functionally transparent and limits its file system impact to a dedicated directory for documentation output. No signs of obfuscation, remote code execution, or credential theft were found.

requirements-clarity