alignment-checkpoint

To require explicit user approval before executing high-stakes code changes, you can use an intent-alignment gate that extracts goals, assumptions, and criteria, then requires a /proceed command before running any tools. This prevents costly mistakes by ensuring no tools run prior to user confirmation.

An intent-alignment gate is a safety mechanism that extracts your explicit goal, implicit assumptions, and success criteria before acting. It prevents misunderstandings by presenting clear options and requiring explicit approval, ensuring the system only executes actions after confirming the exact objective.

Yes, you can preserve a user request verbatim without implementing code changes by using stasis mode. This mode handles preservation-only requests, ensuring the system maintains the exact user input without executing any tools or making system modifications.

To extract assumptions and success criteria from an ambiguous system command request, apply an alignment checkpoint. This process evaluates the request to identify the explicit goal, implicit assumptions, and required criteria, holding further action until a clear go-ahead is provided.

You should use a safety gate for complex system commands when making significant code changes or running potentially risky actions. It acts as an approval gate, confirming the exact objective and preventing tools from running until you explicitly issue a /proceed command.

No, the alignment-checkpoint Skill does not support running tools before user confirmation. It ensures no tools are run prior to explicit /proceed approval, specifically preventing misunderstandings and costly mistakes by holding the system in stasis mode until the user provides a clear go-ahead.