aws-agentic-ai

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

• Potentially malicious external URL detected (high risk: 0.90). The skill instructs runtime operations that fetch external artifacts: the Gateway example references an OpenAPI schema at s3://<BUCKET_NAME>/schemas/my-api-openapi.yaml (fetched to configure tools that affect agent behavior) and the Runtime create command uses a container image "<ACCOUNT_ID>.dkr.ecr..amazonaws.com/my-agent:latest" (pulled and executed as runtime code), both of which are fetched at runtime and can control agent instructions or execute code.