fit-scorer

Pass

Audited by Gen Agent Trust Hub on Aug 20, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes a local Python script youtube.py located within its own package directory to retrieve engagement and authenticity metrics for YouTube influencers. This is a standard functional component provided by the author to automate data retrieval.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests external data from social platforms and user-provided influencer handles, which represents an indirect prompt injection surface.
  • Ingestion points: Influencer handles, profile data, and engagement metrics fetched from YouTube via a connector script.
  • Boundary markers: The instructions mandate using the structured templates defined in references/scoring-templates.md, which helps separate external data from agent instructions.
  • Capability inventory: The skill uses a subprocess call to run youtube.py and writes scoring reports to the local file system at memory/influencer/fit-scorer/.
  • Sanitization: The skill does not explicitly mention sanitization of the influencer-provided text data before inclusion in reports.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 20, 2026, 07:29 PM
Security Audit — agent-trust-hub — fit-scorer