senior-data-engineer
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DYNAMIC_EXECUTION]: The PipelineGenerator classes in scripts/pipeline_orchestrator.py use the compile() function to validate the syntax of generated Airflow, Prefect, and Dagster code.\n
- Evidence: The validate method in the AirflowGenerator (line 218), PrefectGenerator (line 305), and DagsterGenerator (line 381) classes calls compile(code, '', 'exec') on strings constructed from user-provided configuration data.\n- [INDIRECT_PROMPT_INJECTION]: The skill's tools ingest untrusted external data and configuration files, which could be used to influence the agent's behavior or cause unexpected execution during the code generation or validation process.\n
- Ingestion points: The DataLoader.load method in scripts/data_quality_validator.py (lines 697-740) reads CSV, JSON, and JSONL files. The SQLParser.parse method in scripts/etl_performance_optimizer.py (lines 699-705) processes SQL files or inline queries. The main function in scripts/pipeline_orchestrator.py (line 580) reads pipeline configuration files.\n
- Boundary markers: Absent. The tools do not use specific delimiting or safety instructions when processing external content.\n
- Capability inventory: The pipeline_orchestrator.py script possesses dynamic compilation capabilities (line 218) and generates executable Python code.\n
- Sanitization: Absent. The tools rely on standard parsers (json.load, yaml.safe_load) and regex without specific validation to prevent injection into the generated code templates.
Audit Metadata