research-online
Warn
Audited by Snyk on Feb 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: 1.00). The SKILL.md workflow (Step 4: Spawn Agents) and the detailed agent prompts in references/agent-prompts.md explicitly require using WebSearch/WebFetch on public sites (e.g., github.com, stackoverflow.com, blogs and general web) and to read and synthesize those user-generated/untrusted sources, which can directly influence recommendations and next actions.
MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).
- Potentially malicious external URL detected (high risk: 0.80). The skill spawns agents that perform WebSearch/WebFetch at runtime (e.g., site:github.com / https://github.com and top web results) and then injects those fetched pages' content and metadata into the model's context to drive its prompts and final synthesis, so external content fetched at runtime can directly control agent output.
Audit Metadata