The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes local Python scripts (carousel_scan.py, carousel_leaderboard.py, account_stats.py, etc.) using the python3 command. These scripts are passed arguments such as search queries, profile handles, and URLs which are either provided by the user or discovered during the scraping process.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources that could contain malicious instructions designed to influence the agent's behavior.
Ingestion points: Untrusted content is ingested from TikTok profile bios, captions, and links via the Apify scraper, as well as from the autovirality.com website.
Boundary markers: The instructions lack explicit boundary markers or warnings to the agent to ignore instructions embedded within the scraped data (e.g., in bios or captions).
Capability inventory: The agent has the capability to execute local Python scripts and perform network operations to fetch data from the web.
Sanitization: There is no evidence of sanitization or filtering of the retrieved social media text before it is processed by the agent.
[EXTERNAL_DOWNLOADS]: The skill fetches content from autovirality.com to identify candidate accounts. This involves making network requests to an external domain that is not recognized as a standard service provider.

carousel-benchmark