duckdb-analytics

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt includes examples that hardcode S3 access_key/secret_key and constructs SQL/command strings by interpolating secret values (e.g., SET s3_access_key_id = '{access_key}'), which requires the agent to emit secrets verbatim and is therefore insecure.

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

• Third-party content exposure detected (high risk: 0.80). The SKILL.md explicitly instructs the agent to read Parquet files from S3/MinIO (e.g., read_parquet('s3://data/...') and configure_s3(...) in the Python/SQL examples), which means the agent will fetch and interpret arbitrary third-party bucket data as part of its workflow and that data can materially affect queries and downstream actions.