build-data-pipeline

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

• Third-party content exposure detected (high risk: 0.80). The skill's required workflow explicitly instructs ingesting data from external cloud storage and APIs (e.g., Step 2 "Ingest Raw Data" examples using read_parquet('s3://bucket/...') and the "Source: External data -- files, cloud storage (S3, GCS, Azure), APIs, databases" guidance) and to "explore the live environment," which means the agent will fetch and interpret untrusted/third-party user-provided content that can directly influence pipeline decisions and actions.