pytorch-lightning

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

• Third-party content exposure detected (high risk: 0.80). The skill's DataModule documentation and examples (references/data_module.md and prepare_data() example) explicitly show downloading datasets from arbitrary HTTP URLs (e.g., download_dataset("http://example.com/data.zip")), meaning the runtime workflow can fetch and ingest untrusted public/user-provided content for the agent to read and process.