power-user-cultivation

Pass

Audited by Gen Agent Trust Hub on Mar 5, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill identifies and analyzes external data sources such as social media mentions, community forum posts, and user-generated blog content. This creates a surface for indirect prompt injection where an attacker could embed instructions in the content being monitored.
  • Ingestion points: Data gathering phase in SKILL.md (community participation data, social mentions, blog posts).
  • Boundary markers: None specified in the instructions for the gathered data.
  • Capability inventory: The skill consists of instructions and templates; it does not implement subprocess calls, file-write operations, or network requests.
  • Sanitization: No sanitization or validation of the external content is prescribed.
  • [EXTERNAL_DOWNLOADS]: The skill references well-known external services like Octolens and FirstPromoter for social monitoring and referral management. These are standard industry tools and do not involve the execution of untrusted remote scripts.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 5, 2026, 08:03 PM