The Agent Skills Directory

[SAFE]: The skill provides instructional content and utility scripts for medical device risk management per ISO 14971. No security issues were detected.
[COMMAND_EXECUTION]: The skill includes two Python scripts: scripts/fmea_analyzer.py and scripts/risk_matrix_calculator.py. These scripts are legitimate utilities for calculating risk metrics (RPN and risk levels). They use standard Python libraries (argparse, json, dataclasses, typing, enum, datetime) and do not perform any sensitive operations such as network access, unauthorized file system modification, or privilege escalation.
[INDIRECT_PROMPT_INJECTION]: The scripts/fmea_analyzer.py script identifies a vulnerability surface by ingesting external data via JSON files (e.g., fmea_input.json).
Ingestion points: Data enters the agent context via the --data parameter in scripts/fmea_analyzer.py.
Boundary markers: Absent in script output.
Capability inventory: The skill facilitates command execution of its own scripts and file reading across its reference directory.
Sanitization: The scripts perform type conversion (e.g., to integers for RPN calculation) but do not sanitize descriptive text fields (e.g., failure_mode). This is a standard risk for data analysis tools and is considered a low-severity finding in the context of its intended use.

risk-management-specialist