pymc-bayesian-modeling

Warn

Audited by Gen Agent Trust Hub on May 6, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [DYNAMIC_EXECUTION]: The reference file references/advanced_workflows.md provides code examples that use the pickle module to load data.
  • Evidence: with open("model.pkl", "rb") as f: saved = pickle.load(f) in the 'Save/Load Patterns' section.
  • Risk: The pickle.load() function is vulnerable to arbitrary code execution if it processes a file from an untrusted source.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external data without using delimiters or instructions to ignore embedded commands.
  • Ingestion points: Data entry points via NumPy arrays and Pandas DataFrames in SKILL.md.
  • Boundary markers: Absent; no markers or instructions are provided to separate user data from agent instructions.
  • Capability inventory: The skill enables the agent to execute complex Python code, install packages, and read/write to the filesystem.
  • Sanitization: Absent; input data is used directly in modeling contexts.
  • [EXTERNAL_DOWNLOADS]: The skill instructs the user and agent to install third-party libraries via a package manager.
  • Evidence: pip install pymc arviz numpy matplotlib in the 'Prerequisites' section of SKILL.md.
  • [COMMAND_EXECUTION]: The skill provides numerous Python code snippets for the agent to execute in its environment to build and validate statistical models.
Audit Metadata
Risk Level
MEDIUM
Analyzed
May 6, 2026, 01:46 AM
Security Audit — agent-trust-hub — pymc-bayesian-modeling