clean-code

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFE
Full Analysis
  • [Indirect Prompt Injection] (SAFE): The skill processes untrusted code which could contain malicious instructions. However, the risk is mitigated by the skill's restricted execution environment.
  • Ingestion points: The 'code' parameter in SKILL.md is the primary entry point for external data.
  • Boundary markers: No explicit delimiters are used in the prompt instructions.
  • Capability inventory: The skill is atomic and idempotent, with no access to network, file-system writing, or shell execution.
  • Sanitization: No input sanitization is performed.
  • [Data Exposure] (SAFE): No hardcoded credentials, API keys, or sensitive file paths were detected. The Python validation script correctly uses 'yaml.safe_load()' to prevent unsafe deserialization.
  • [Remote Code Execution] (SAFE): No remote code execution patterns, such as piping downloads to a shell, were found. The skill relies on local metrics calculation.
  • [Obfuscation] (SAFE): No evidence of Base64, zero-width characters, or other obfuscation techniques was found in the skill's files or metadata.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:36 PM