The Agent Skills Directory

[SAFE]: The skill implements checkpoint and rollback management. While these are destructive operations (state restoration and resets), the skill explicitly enforces safety protocols, specifically the requirement for user confirmation via AskUserQuestion regardless of the agent's automation level. It also creates safety checkpoints before performing any rollback or reset. All file operations target internal state directories (.bkit/checkpoints/, .bkit/state/) and local documentation folders, with no evidence of remote data exfiltration, unauthorized privilege escalation, or obfuscation. Dependencies are local JavaScript modules for state management and auditing.

rollback