book-market-research

Warn

Audited by Snyk on Apr 21, 2026

Risk Level: MEDIUM
Full Analysis

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

  • Third-party content exposure detected (high risk: 0.90). The SKILL.md and references/research-methods.md explicitly state Claude performs qualitative research "via web search" and pulls review sources like Goodreads and Reddit (user-generated, public sites) to identify competitor positioning and review themes that directly influence the viability score and recommendations, so the agent will ingest untrusted third‑party content as part of its core workflow.

Issues (1)

W011
MEDIUM

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

Audit Metadata
Risk Level
MEDIUM
Analyzed
Apr 21, 2026, 05:52 PM
Issues
1