The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill contains an attack surface for indirect prompt injection as its primary function is to ingest and analyze untrusted external data from App Store reviews and CSV files. This is handled through a structured analysis framework.
Ingestion points: Reviews are accessed via the get_review_data and get_reviews_data tools, as well as user-uploaded CSV files.
Boundary markers: There are no explicit delimiters or markers defined to separate the untrusted review content from the agent's instructions.
Capability inventory: The skill has the ability to draft and apply responses back to the App Store using fill_review and fill_reviews tools.
Sanitization: No specific sanitization or filtering steps are documented for the review content before it is processed by the AI.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill includes positive security practices and explicit constraints to prevent data leaks. It instructs the agent to never reveal internal data and to avoid asking for personal information in public responses.

reviews-ratings