NYC

color-theory-palette-harmony-expert

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFE
Full Analysis
  • External Dependencies (SAFE): The skill suggests standard, reputable Python packages (colormath, opencv-python, numpy, scipy, scikit-image, pot, hnswlib, scikit-learn) essential for the described color science tasks. No suspicious or unversioned remote script execution was found.
  • Indirect Prompt Injection Surface (LOW): The skill utilizes mcp__firecrawl__firecrawl_search and WebFetch to ingest untrusted data from the web. When paired with high-privilege tools like Bash, Write, and Edit, this creates a theoretical vulnerability surface.
  • Ingestion points: Firecrawl, WebFetch (documented in SKILL.md).
  • Boundary markers: Absent; the instructions do not explicitly tell the agent to ignore instructions embedded in fetched research papers or web content.
  • Capability inventory: Bash, Write, Edit, Stability AI (full control over local environment).
  • Sanitization: None specified for external content.
  • Technical Integrity (SAFE): The documentation contains high-quality, accurate code snippets for color space transformations and optimal transport algorithms. There are no signs of obfuscation, hidden instructions, or credential harvesting.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 05:24 PM