skill-model-updater

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

• Potentially malicious external URL detected (high risk: 0.90). This skill explicitly performs a runtime WebFetch of https://docs.anthropic.com/en/docs/about-claude/models to discover current model IDs and then injects those discovered values into the updater's actions (deciding and writing model strings), meaning the fetched external content directly controls the agent's behavior.