harness2

Warn

Audited by Snyk on Nov 11, 2026

Risk Level: MEDIUM
Full Analysis

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

  • Third-party content exposure detected (low risk: 0.10). The required benchmark runs let executor rollouts reach open network access, where they can fetch arbitrary web content (including public rubrics), so outsider-authored text can be read without a specific item being chosen first; the skill's own network-isolation guidance acknowledges this exposure.

MEDIUM W013: Attempt to modify system services in skill instructions.

  • Attempt to modify system services in skill instructions detected (low risk: 0.30). The skill mostly gates its machine-changing steps (git clones, installs, Docker pulls, server start) behind explicit approval, but step 5 asks the agent to block domains at the network layer, which modifies host network configuration and may need elevated privileges, and it never states that sudo is off-limits.

Issues (2)

W011
MEDIUM

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

W013
MEDIUM

Attempt to modify system services in skill instructions.

Audit Metadata
Risk Level
MEDIUM
Analyzed
Nov 11, 2026, 08:30 AM
Issues
2