python-security

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly instructs the agent to "Load" external standard documents at runtime (e.g., ../python-core/standards/python-core.md and ../../../plan-marshall/skills/persona-security-expert/standards/input-validation-trust-boundaries.md), meaning fetched content will directly control the agent's guidance and behavior, so these runtime-loaded relative paths are flagged.