check-architecture-principles

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly fetches principle documents at runtime via the GitHub contents API (e.g., gh api repos/camunda/product-development/contents/architecture/principles and repos/camunda/product-development/contents/architecture/principles/{filename}) and/or a fallback "github" MCP server, and those fetched documents are read and injected into the agent's evaluation logic, so remote content directly controls the agent's prompts.