The Agent Skills Directory

[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection due to its processing of untrusted external web content.
Ingestion points: The skill retrieves full App Store and Google Play listings, including user-controlled text like descriptions and reviews, using WebFetch and search queries as described in SKILL.md.
Boundary markers: There are no instructions in SKILL.md to use specific delimiters (such as XML tags or triple quotes) or to include "ignore embedded instructions" warnings when the agent processes the fetched listing data.
Capability inventory: The skill utilizes WebFetch for data retrieval and references taking screenshots of the listing pages to analyze visual assets, providing a surface for visual or text-based injection.
Sanitization: The instructions lack any requirement for the agent to filter, sanitize, or validate the content retrieved from external URLs before analyzing it against the scoring criteria.

aso