social-trend-monitor
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- Prompt Injection (LOW): Potential for indirect prompt injection via external social media data.
- Ingestion points: Public posts, titles, and descriptions retrieved from Reddit, Instagram, and TikTok via WebSearch and WebFetch.
- Boundary markers: Absent. The skill lacks instructions to treat retrieved data as untrusted or to ignore embedded commands.
- Capability inventory: Uses the Write tool to save files to ~/.claude/cache/social-trends/ and WebSearch/WebFetch tools for network access.
- Sanitization: Absent. Content is directly formatted into a report without escaping or filtering.
Audit Metadata