The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses CronCreate and CronDelete to automate recurring tasks. It executes commands and spawns sub-agents via the Bash and Agent tools to perform code refactoring and testing based on user-defined goals.
[SAFE]: The 'Blind Review Principle' described in the skill is a security and quality best practice. It explicitly instructs the dispatcher to strip historical context (prior failures, fix details) when spawning review agents, forcing them to re-evaluate the code from scratch without bias.
[SAFE]: All tool usage (Read, Write, Edit, Glob, Grep, Bash, Agent) is scoped to the allowed-tools frontmatter and follows the intended purpose of the skill for codebase refactoring.
[SAFE]: The skill implements a 'Sample' phase to ensure quality persistence by requiring a specific number of consecutive passing checks before auto-stopping the schedule.

powerloop