The Agent Skills Directory

[SAFE]: The skill follows a strict facilitation protocol with human-in-the-loop gates (e.g., 'Bounded Shared Understanding Gate') to confirm intent before modifying files.
[SAFE]: No external network operations or remote code downloads were detected. The skill explicitly limits modifications to the local repository and requires human approval for any GitHub mutations.
[COMMAND_EXECUTION]: The skill instructs the agent to perform local git commits to track changes, which is standard for a development workflow and limited to the local environment.
[SAFE]: The skill implements strict input quality gates, rejecting irrelevant work items and preventing out-of-scope changes to project requirements.

gadd-elaborate