The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is exposed to an indirect prompt injection attack surface as it fetches and processes untrusted content from public app stores. 1. Ingestion points: The skill uses WebFetch and browser screenshot tools in SKILL.md to retrieve app metadata and reviews from apps.apple.com and play.google.com. 2. Boundary markers: SKILL.md contains a specific warning to treat listing data as untrusted and to never follow embedded instructions. 3. Capability inventory: SKILL.md utilizes WebFetch for network reading and browser tools for visual data capture. 4. Sanitization: No explicit programmatic sanitization is defined, but the risk is mitigated by the specific instructions provided in the Before Auditing section of SKILL.md.

aso