customer-research

Pass

Audited by Gen Agent Trust Hub on Jul 3, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface due to its core function of processing untrusted external data.
  • Ingestion points: The skill ingests data from external public sources such as Reddit comments (reddit/posts/{id}/comments), TikTok comments (tiktok/videos/{id}/comments), and news articles via UnifAPI.
  • Boundary markers: There are no explicit instructions for the agent to use delimiters or ignore potential instructions embedded within the ingested social media comments.
  • Capability inventory: The skill is primarily designed for data retrieval and report generation; it does not explicitly request file-write or arbitrary code execution capabilities, which limits the potential impact of an injection.
  • Sanitization: The instructions do not specify any automated sanitization or filtering of the retrieved content, although they do emphasize verbatim quoting and citation.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 3, 2026, 04:06 PM
Security Audit — agent-trust-hub — customer-research