vaex

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly loads external/untrusted data and state from arbitrary file/URI sources (e.g., vaex.open('s3://bucket-name/data.parquet'), vaex.open('gs://...'), vaex.open('ws://hostname:9000/data'), vaex.from_csv(...), and df.state_load('state.json')), so the agent would ingest and interpret third-party/user-provided content as part of its workflow.