eve-job-debugging

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • Prompt Injection (SAFE): No instructions targeting safety bypass, role-play, or behavior override were detected.\n- Indirect Prompt Injection (SAFE): The skill defines an ingestion surface for untrusted data via the processing of eve CLI tool outputs. However, the skill does not possess high-risk capabilities (like file write or network exfiltration) that could be exploited through this surface.\n
  • Ingestion points: eve job follow, eve job wait, eve job result, eve job diagnose, eve job show, eve job dep list.\n
  • Boundary markers: Absent.\n
  • Capability inventory: Shell execution of specifically documented eve commands.\n
  • Sanitization: Absent.\n- No Code (SAFE): No Python packages, Node.js modules, shell scripts, or binary executables are included in the skill. It is strictly instructional documentation.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 05:21 PM