The Agent Skills Directory

[SAFE]: No malicious patterns detected. The provided Python scripts (commercial_policy_generator.py, policy_compliance_checker.py, terms_deviation_analyzer.py) use only the Python standard library and do not perform network operations or access sensitive system files.
[SAFE]: The skill uses a custom, manual YAML parser in its scripts rather than potentially unsafe loading libraries, which follows security best practices for data ingestion.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface (Category 8) as it processes deal data from external CSV files to generate reports for the agent. However, the risk is minimal as the scripts perform strict type casting (int, float) on numerical data and do not interpret instructions from the data source.
Ingestion points: CSV data read via the --deals flag in scripts/policy_compliance_checker.py and scripts/terms_deviation_analyzer.py.
Boundary markers: None present in the generated markdown output; however, the agent is directed to use the tool for specific analytical tasks.
Capability inventory: File system read/write is limited to the user-provided deal and policy files.
Sanitization: Data validation is performed through explicit casting to float and integer types for policy thresholds and deal values.

commercial-policy