senior-data-scientist
Pass
Audited by Gen Agent Trust Hub on Aug 21, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill consists of informational markdown documentation and boilerplate Python scripts. No security risks were identified during the analysis of the code or instructions.
- [INDIRECT_PROMPT_INJECTION]: The scripts in the scripts/ directory accept input file paths via CLI arguments, establishing a surface for indirect prompt injection if malicious data is processed. However, the current code is skeleton-only with no active data handling. * Ingestion points: --input argument in scripts/experiment_designer.py, scripts/feature_engineering_pipeline.py, and scripts/model_evaluation_suite.py. * Boundary markers: Absent. * Capability inventory: Basic file path handling; no network or shell execution. * Sanitization: Not implemented.
- [METADATA_POISONING]: There are minor inconsistencies between the command examples in SKILL.md and the actual argument parsing logic in the provided Python scripts (e.g., missing --target and --deploy flags). This appears to be a documentation oversight rather than a malicious attempt to deceive.
Audit Metadata