meta-apply

To apply patches to a skill corpus safely, use a privileged execution process that requires cross-model jury review and explicit human approval before any changes land, ensuring structural integrity is maintained.

Cross-model jury review for patch application is a validation mechanism where multiple models impartially assess proposed modifications before they are applied, ensuring no single loop can apply its own patch.

Yes, you can apply all staged patches to the skill corpus at once by passing the 'all' argument to the execution command, provided every patch has passed fresh cross-model jury review and human approval.

Patch application requires human approval to guarantee that privileged actions modifying the skill corpus are explicitly authorized, preventing unauthorized or automated loops from compromising structural integrity.

Meta-optimize patches land without self-application loops by enforcing a strict separation where no loop can apply its own patch, requiring fresh cross-model jury review at landing time and human authorization.