Human in the loop

What

The human owns authorization, design trade-offs, execution permissions, review, and merging.

Why

AI proposes; it does not assume accountability.

When

Always, especially around data, security, cost, and public contracts.

Example

Require an explicit human decision in REVIEW.md.

Common mistake

Giving an agent broad permissions because a task appears routine.

Checklist

• [ ] Permissions are least-privilege
• [ ] Output is independently reviewed

Next step

Choose an adapter.