professor-critic

Pass

Audited by Gen Agent Trust Hub on Jul 9, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [PROMPT_INJECTION]: No malicious override patterns or safety bypass instructions were found. The assertive language used is intended solely to establish a rigorous critique persona.
  • [DATA_EXFILTRATION]: No network commands, hardcoded credentials, or sensitive file path access patterns were identified in the skill instructions.
  • [REMOTE_CODE_EXECUTION]: The skill does not contain any code, external dependencies, or commands that facilitate remote code execution.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-provided artifacts (text, proofs, code) for review. Ingestion points: Artifact intake defined in SKILL.md. Boundary markers: Absent. Capability inventory: Text generation only (no tool access). Sanitization: Absent. The security risk is assessed as safe because the skill has no access to sensitive capabilities or the host system.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 9, 2026, 02:50 PM
Security Audit — agent-trust-hub — professor-critic