alterlab-pymc
Warn
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: MEDIUMDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DYNAMIC_EXECUTION]: The skill provides documentation and script templates that utilize the Python 'pickle' module for serializing and deserializing PyMC models and results. Specifically, 'references/workflows.md' and 'scripts/model_comparison.py' contain examples of 'pickle.dump()' and 'pickle.load()'. Deserializing data with 'pickle.load()' from an untrusted source is a known vulnerability that can lead to arbitrary code execution.
- [INDIRECT_PROMPT_INJECTION]: This skill possesses an attack surface for indirect prompt injection because it is designed to ingest and process external datasets for statistical modeling.
- Ingestion points: Data loading steps in 'assets/hierarchical_model_template.py' and 'assets/linear_regression_template.py' indicate where untrusted external data (e.g., CSV files) enters the agent's context.
- Boundary markers: The instructions do not specify the use of delimiters or warnings to ignore instructions embedded within the data.
- Capability inventory: The skill utilizes file writing capabilities (saving plots, NetCDF results, and CSV summaries) and execution capabilities via the 'Bash' tool for running Python and 'uv'.
- Sanitization: There is no evidence of validation or sanitization for the content of the data files processed by the modeling scripts.
Audit Metadata