deep-research
Warn
Audited by Snyk on Aug 14, 2026
Risk Level: MEDIUM
Full Analysis
MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).
- Third-party content exposure detected (medium risk: 0.30). In SKILL.md, the runtime workflow explicitly ingests outsider-authored free text by running
/last30days(Reddit/X/YouTube/HN/web recency) and summarizing captured results, which can include attacker-submitted poison text without requiring the agent to pre-select a specific item.
Issues (1)
W011
MEDIUMThird-party content exposure detected (indirect prompt injection risk).
Audit Metadata