The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data from the App Store, creating a potential surface for indirect prompt injection.\n
Ingestion points: The store-signals/SKILL.md sub-skill retrieves user reviews via the mcp__asc-metadata__list_reviews tool, and store-growth-audit/SKILL.md reads app metadata.\n
Boundary markers: No explicit delimiters or warnings to ignore embedded instructions are used when processing reviews or metadata.\n
Capability inventory: The skill is authorized to use Write for project planning files (e.g., ROADMAP.md, SIGNALS.md) and WebSearch for external queries.\n
Sanitization: There is no evidence of sanitization or filtering for the retrieved review content before it is processed or written to disk.

growth