github-deep-research

Warn

Audited by Snyk on Apr 28, 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). Yes — the SKILL.md research workflow (Rounds 2–3) explicitly instructs using web_search and web_fetch plus the GitHub API to ingest public, user-generated sources (repo READMEs, issues/PRs, Reddit/HN/social media) which the agent must read and use to drive analysis and actions, exposing it to untrusted third‑party content that could contain injected instructions.

Issues (1)

W011
MEDIUM

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

Audit Metadata
Risk Level
MEDIUM
Analyzed
Apr 28, 2026, 01:27 AM
Issues
1
Security Audit — snyk — github-deep-research