clinical-decision-support

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.85). The skill contains explicit, mandatory integration with an external AI image-generation/review service that encodes and uploads document content (including images) to a third‑party API, creating a high risk of unintended data exfiltration (including PHI) and use of base64-encoded payloads for external execution/review; no hidden remote shells or obfuscated execs were found, but the external upload flow and mandatory figure generation make this a high-risk data‑exfiltration pattern.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill's runtime scripts (e.g., scripts/generate_schematic_ai.py and scripts/generate_schematic.py) make live API calls to https://openrouter.ai/api/v1 (OpenRouter/Nano Banana models) and use the model's responses to review, modify, and control subsequent prompt generation and image-generation behavior, so an external URL is contacted at runtime and its returned content directly controls agent prompts/instructions.

MEDIUM W021: Hidden or invisible Unicode characters detected (potential obfuscation or prompt injection).

• Hidden Unicode characters detected (1 type(s) found)