The Agent Skills Directory

[PROMPT_INJECTION]: The skill documentation in SKILL.md uses imperative language ('CRITICAL: Anti-Loop Guardrails') to define constraints that override other instructions. These are designed as stability measures to prevent the agent from entering infinite self-improvement loops rather than attempting to bypass safety filters.
[COMMAND_EXECUTION]: The skill includes several automation scripts. scripts/activator.sh and scripts/error-detector.sh are intended for use in agent hooks to provide contextual reminders. The scripts/extract-skill.sh utility allows for scaffolding new skills and implements strict validation on input parameters (enforcing alphanumeric naming and preventing path traversal) to ensure file operations remain within the designated workspace.
[EXTERNAL_DOWNLOADS]: Installation instructions in README.md suggest cloning from a GitHub repository. This is a standard distribution method using a well-known service and aligns with the expected behavior for a skill of this nature.

self-improvement