extract-fuzzer-repro

Warn

Audited by Snyk on Feb 17, 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 instructs the agent to fetch and parse GitHub Actions logs from a public repository (using "gh api repos/noir-lang/noir/actions/jobs/JOB_ID/logs") and to read reproduced code from those logs, which are untrusted, user-generated third-party content that the agent will interpret as part of its workflow.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 17, 2026, 12:24 AM