The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains a vulnerability surface for indirect prompt injection via the processing of external App Store reviews.
Ingestion points: The script scripts/app_review_watch.sh (lines 25 and 36) fetches user-generated content from Apple's RSS feeds (itunes.apple.com).
Boundary markers: The skill does not employ delimiters or "ignore instructions" markers when echoing review content into the agent's context.
Capability inventory: The agent is granted broad capabilities via the Bash, Read, and Write tools, which could be targeted by instructions hidden in reviews.
Sanitization: No validation or sanitization is performed on the incoming review text to prevent the execution of embedded commands.
[COMMAND_EXECUTION]: The skill utilizes dynamic script generation and execution patterns, specifically Python and Ruby heredocs (python3 << 'EOF'), to manage JWT signing and complex API requests.
[SAFE]: All external communications are directed to trusted domains belonging to Apple and the well-known service RevenueCat. The handling of sensitive files like .p8 private keys and .env configuration is consistent with the primary purpose of managing App Store services.

apple-appstore-manager