The Agent Skills Directory

[SAFE]: The skill provides comprehensive documentation and templates referencing administrative system commands such as dd, tc, iptables, and kill. These are correctly contextualized as standard tools for chaos engineering experiments, such as simulating disk exhaustion or network partitions, and do not represent an autonomous execution risk by the agent.
[SAFE]: The provided Python scripts (blast_radius_calculator.py, chaos_experiment_designer.py, and gameday_planner.py) are benign. They rely solely on the Python standard library for performing calculations and generating markdown artifacts without any network activity or unauthorized file system modifications.
[SAFE]: Although the skill includes a surface for indirect prompt injection by interpolating user-supplied arguments (e.g., service names) into generated markdown templates, this behavior is essential to its function as a scaffolding tool. The risk is considered minimal and appropriate for its intended use case.
[SAFE]: No evidence of malicious obfuscation, hidden credentials, or unauthorized data exfiltration was found during the analysis. The skill's behavior is transparent and consistent with its stated engineering objectives.

chaos-engineering