pymc-bayesian-modeling

Warn

Audited by Gen Agent Trust Hub on Mar 3, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [DYNAMIC_EXECUTION]: The documentation includes code snippets demonstrating the use of the Python 'pickle' module for serializing and deserializing PyMC models and results.
  • Evidence: In references/workflows.md, the code sample shows with open('model.pkl', 'rb') as f: saved = pickle.load(f). Deserializing untrusted data with pickle can lead to arbitrary command execution on the host system.
  • [INDIRECT_PROMPT_INJECTION]: The skill defines workflows that ingest external data, creating an attack surface where malicious instructions could be embedded in the data to influence agent behavior.
  • Ingestion points: Data is ingested via pd.read_csv('data.csv') in SKILL.md and assets/hierarchical_model_template.py.
  • Boundary markers: No boundary markers or instructions to ignore embedded content are used during data ingestion.
  • Capability inventory: The skill possesses file-writing capabilities through idata.to_netcdf(), summary.to_csv(), and plt.savefig() across multiple scripts and templates.
  • Sanitization: There is no evidence of data validation, sanitization, or schema enforcement before the data is processed by the modeling engine.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 3, 2026, 08:47 PM