The Agent Skills Directory

[SAFE]: The skill was comprehensively audited for malicious patterns including data exfiltration, persistence mechanisms, and privilege escalation; no such behaviors were identified.
[COMMAND_EXECUTION]: The framework uses subprocess calls to execute its own Python logic and standard git commands. It also employs dynamic context injection (!command syntax) to display status information. All such executions are restricted to safe, local, and hardcoded paths and are used exclusively for coordination and state reporting.
[PROMPT_INJECTION]: The instructions and templates were analyzed for prompt injection and behavioral overrides; no malicious patterns, safety bypass attempts, or extraction triggers were found.

nelson