social-media-analyzer

Pass

Audited by Gen Agent Trust Hub on Feb 19, 2026

Risk Level: SAFE
Full Analysis
  • PROMPT_INJECTION (SAFE): No behavioral overrides or safety bypass markers were detected in the skill instructions.
  • DATA_EXFILTRATION (SAFE): No hardcoded credentials, sensitive file paths, or network operations were found.
  • COMMAND_EXECUTION (SAFE): The skill uses local scripts for metric calculations; these scripts contain no dangerous functions like eval() or subprocess calls with untrusted input.
  • INDIRECT_PROMPT_INJECTION (SAFE): [Ingestion points]: Campaign data in assets/sample_input.json. [Boundary markers]: None. [Capability inventory]: Local metrics calculation with no network or write-access. [Sanitization]: None. The finding is marked as SAFE as the data ingestion surface is essential for the skill's primary purpose and lacks executable capabilities.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 19, 2026, 01:51 AM