cro-methodology

Warn

Audited by Snyk on Feb 19, 2026

Risk Level: MEDIUM
Full Analysis

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

  • Third-party content exposure detected (high risk: 0.80). Yes — SKILL.md (see "Gather Market Intelligence") and the referenced docs (OBJECTIONS.md, RESEARCH.md) explicitly instruct monitoring and analyzing social media, public reviews and forums (e.g., "Reddit, Twitter threads", "Reviews") — i.e., untrusted, user-generated third‑party content the agent would fetch/ingest and that can materially influence its CRO decisions and next actions.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 19, 2026, 03:51 PM