The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's documentation and workflows (e.g., in SKILL.md and references/workflows.md) recommend using joblib.load() and shap.Explainer.load() to restore models and explainer objects. These methods typically use the pickle serialization format, which is prone to arbitrary code execution if the loaded file is from an untrusted source.\n- [PROMPT_INJECTION]: The skill's primary function is to ingest and analyze external datasets and model outputs, which establishes a surface for indirect prompt injection.
Ingestion points: Testing datasets (X_test) and model objects provided for interpretation.
Boundary markers: The skill lacks explicit markers or safety instructions to prevent the agent from interpreting embedded instructions within the processed data.
Capability inventory: The skill demonstrates file system access through joblib.dump and matplotlib.pyplot.savefig.
Sanitization: No input validation or data sanitization routines are described in the provided code examples.

shap