clip

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADS
Full Analysis
  • [EXTERNAL_DOWNLOADS] (SAFE): The skill recommends installing the CLIP library directly from OpenAI's official GitHub repository.
  • Evidence: pip install git+https://github.com/openai/CLIP.git in SKILL.md.
  • Trust Status: The openai organization is a recognized trusted source; under [TRUST-SCOPE-RULE], this download is considered safe.
  • [INDIRECT_PROMPT_INJECTION] (SAFE): While the skill processes external images and text which constitutes a potential attack surface for adversarial inputs, this is inherent to the primary purpose of vision-language models and does not represent a specific vulnerability in the skill's implementation.
  • Ingestion Points: Image.open("photo.jpg") and clip.tokenize() in SKILL.md and references/applications.md.
  • Capability Inventory: Standard PyTorch inference and vector database (Chroma) operations.
  • Sanitization: The skill uses standard preprocessing functions provided by the CLIP library.
  • [COMMAND_EXECUTION] (SAFE): The provided code examples use standard library calls and established ML frameworks without invoking dangerous shell commands or arbitrary execution logic.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:06 PM