ml-engineer

Warn

Audited by Gen Agent Trust Hub on Feb 22, 2026

Risk Level: MEDIUMSAFE
Full Analysis
  • Dynamic Execution (MEDIUM): The skill recommends the 'Pickle' format for model deployment and serialization.
  • Evidence: The instructions in 'SKILL.md' state to 'Export models to standard formats (ONNX, Pickle, SavedModel)'.
  • Risk: The Python 'pickle' module is insecure because it can execute arbitrary code during the deserialization process. Loading an untrusted pickle file can lead to host compromise.
  • Indirect Prompt Injection (LOW): The skill is designed to ingest and process data from external files, which represents a potential attack surface.
  • Ingestion points: The skill reads data from 'data.csv' using 'pandas.read_csv()'.
  • Boundary markers: There are no instructions defining delimiters or boundary markers to prevent the model from interpreting instructions embedded within the data files.
  • Capability inventory: The agent is equipped with the capability to execute Python code for data processing and model training.
  • Sanitization: No sanitization or validation of the input data is suggested before it is processed by the training pipeline.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 22, 2026, 09:36 PM