The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill includes a customMapFunction input parameter that accepts a JavaScript function as a string. This string is intended to be executed at runtime to transform data, which allows for arbitrary code execution within the execution environment of the scraper.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process large amounts of untrusted data from external Instagram profiles, creating a potential attack surface.
Ingestion points: Metadata fields such as biography, fullName, and externalUrl are fetched from public Instagram profiles and returned to the agent context.
Boundary markers: The skill documentation does not specify the use of delimiters or instructions to ignore embedded commands within the scraped data.
Capability inventory: The skill uses node scripts for local execution and curl for network requests to the Apify API.
Sanitization: There is no mention of sanitization, filtering, or escaping for the scraped bio text or other user-controlled fields before they are presented to the AI agent.
[COMMAND_EXECUTION]: The documentation provides shell command templates using node and curl. While these facilitate legitimate tool usage, they involve shell execution and the use of sensitive environment variables like $APIFY_TOKEN.

scraping-instagram-users-by-keyword