ai-sdk

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly instructs runtime fetches of external content—e.g., the model list via curl to https://ai-gateway.vercel.sh/v1/models and docs from https://ai-sdk.dev/...—and relies on those responses to choose models and drive code/prompts, so these URLs directly control agent behavior at runtime.