self-improvement

Pass

Audited by Gen Agent Trust Hub on Mar 11, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill's core functionality involves maintaining a local database of markdown files in a .learnings/ directory. This is a legitimate and transparent use of the agent's file system capabilities for workspace management.
  • [COMMAND_EXECUTION]: Helper scripts such as activator.sh, error-detector.sh, and extract-skill.sh are included for automation. These scripts use standard Linux utilities and implement regex-based input validation for parameters like skill names to prevent command injection.
  • [PROMPT_INJECTION]: The skill defines detection triggers for the agent to recognize when a learning should be logged. These are behavioral guidelines and do not attempt to override safety filters, bypass system prompts, or exfiltrate sensitive data.
  • [EXTERNAL_DOWNLOADS]: All installation instructions and repository references point to the author's official GitHub namespace (pskoett/pskoett-ai-skills), which is consistent with the vendor author context and categorized as safe.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 11, 2026, 05:12 AM