The Agent Skills Directory

[SAFE]: No malicious patterns, command injection, or data exfiltration attempts were detected. The skill is focused on providing structured analysis tools for failure mode identification and risk assessment.
[INDIRECT_PROMPT_INJECTION]: The skill manages a potential indirect prompt injection surface where user-provided FMEA data is processed. 1. Ingestion points: Failure descriptions, effects, causes, and recommended actions collected during the 7-step workflow (SKILL.md). 2. Boundary markers: Includes explicit instructions in SKILL.md ('Input Handling and Content Security') to treat user text as data and ignore embedded instructions. 3. Capability inventory: Local JSON/HTML file operations via scripts (scripts/calculate_fmea.py, scripts/generate_report.py). No network access or dynamic code execution. 4. Sanitization: Employs 'html.escape()' in generate_report.py to prevent XSS in output files.
[COMMAND_EXECUTION]: The provided Python scripts (scripts/calculate_fmea.py, scripts/generate_report.py, scripts/score_analysis.py) are intended for local execution. They incorporate path validation to prevent directory traversal attacks and restrict operations to authorized file extensions, ensuring they operate within the intended environment.

fmea-analysis