lov-create-qrcode

Pass

Audited by Gen Agent Trust Hub on Sep 1, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill performs QR code generation and verification entirely locally using standard Python libraries. No network-based exfiltration or remote resource fetching was identified in the operational scripts.- [DATA_EXPOSURE_AND_EXFILTRATION]: The skill manages user preferences across sessions using a local JSON profile. The scripts/profile_store.py script includes a security filter that explicitly prevents the storage of sensitive fields (e.g., tokens, secrets, passwords) in the durable profile.- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data to generate images, which is its primary function. It mitigates injection risks through explicit developer guidelines.
  • Ingestion points: scripts/create_qrcode.py accepts payloads from stdin, command-line arguments, and local files.
  • Boundary markers: Instructions in SKILL.md mandate the use of stdin for private data and forbid the persistence of sensitive payloads.
  • Capability inventory: The skill uses PIL and qrcode for generation, cv2 for verification, and writes to the local file system.
  • Sanitization: The skill provides structural and scan-based verification to ensure artifact integrity and reports payload metrics (hash/length) in JSON receipts without echoing the raw content.- [DYNAMIC_EXECUTION]: The use of subprocess.run is restricted to the unit test suite (tests/test_create_qrcode.py) to verify CLI functionality. It does not appear in the production logic, posing no risk to the runtime environment.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 1, 2026, 02:44 AM
Security Audit — agent-trust-hub — lov-create-qrcode