package-private-ml-artifacts

Pass

Audited by Gen Agent Trust Hub on Aug 26, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides a legitimate security utility. The script scripts/build_artifact_manifest.py is designed to detect and block the accidental inclusion of sensitive files (like .env) or specific credential strings (Hugging Face tokens, OpenAI keys, and private keys) in ML artifact packages.
  • [SAFE]: No remote code execution or unauthorized network activity was detected. The shell commands provided in SKILL.md invoke the skill's own local scripts using standard Python libraries without external dependencies.
  • [SAFE]: The skill's ingestion of local files for manifest generation creates a surface for potential indirect prompt injection.
  • Ingestion points: scripts/build_artifact_manifest.py (reads artifacts from user-specified directory)
  • Boundary markers: The manifest is generated as a structured JSON file.
  • Capability inventory: File reading, SHA-256 hashing, and writing a JSON manifest.
  • Sanitization: Content is scanned for sensitive regex patterns to prevent exposure. This implementation is considered safe given its defensive purpose.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 26, 2026, 04:00 AM
Security Audit — agent-trust-hub — package-private-ml-artifacts