mlops

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.65). SKILL.md requires the runtime agent to call awsknowledge MCP tools to “read/recommend” AWS documentation, which typically involves fetching and ingesting outsider-authored free-text (public AWS docs) into the agent’s LLM context.

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

• Potentially malicious external URL detected (high risk: 0.70). The CLI training job references and will pull/execute the remote container image 763104351884.dkr.ecr.us-east-1.amazonaws.com/pytorch-training:2.1.0-gpu-py310-cu118-ubuntu20.04-sagemaker at runtime, which is external code executed by the service.