The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the Apple App Store, including user-generated reviews and app descriptions.
Ingestion points: Data enters the agent context via itunes.apple.com and apps.apple.com endpoints (e.g., in endpoints/reviews.md and endpoints/app-lookup.md).
Boundary markers: The provided examples do not use delimiters or explicit instructions to ignore embedded commands within the scraped JSON/HTML content.
Capability inventory: The skill utilizes shell commands like curl, jq, grep, and sed to process data, providing a functional path for further actions based on scraped content.
Sanitization: No evidence of sanitization or filtering of external content (like review text) is present in the provided command-line examples.
[EXTERNAL_DOWNLOADS]: The skill makes network requests to fetch App Store data.
Evidence: Multiple files (e.g., endpoints/search.md, endpoints/lists.md) use curl to reach itunes.apple.com and search.itunes.apple.com. These are official Apple domains and well-known services, which are documented here neutrally.
[COMMAND_EXECUTION]: The skill relies heavily on executing shell commands for its primary function.
Evidence: Examples throughout the skill use curl, jq, grep, sort, head, and tr to retrieve and format data. While these are intended for legitimate scraping, they represent the skill's operational capabilities.

app-store-scraper