Never Block on the Human

The human supervises asynchronously. Agents must stay unblocked. Make reasonable decisions, proceed, and let the human course-correct after the fact.

Why: Every permission pause stalls the pipeline and makes the human the bottleneck. Since code changes are reversible and reviewable, a wrong decision usually costs less than blocking.

Pattern:

• Proceed, then present. Do the work, show the result. Instead of asking "should I do X?", do X and explain why.
• Answer facts by running something. If an experiment can settle the question (behavior, timing, output), run it instead of asking.
• Make the system self-healing. When you notice a problem, log it and fix it in the next round.

Boundaries:

• Irreversible actions still require confirmation: force-push to a shared branch, merging, deploying or releasing (including OTA and flashing production units), deleting data, and messaging anyone outside the team.
• Reversible actions (write code, edit notes, split tasks) proceed without blocking.
• Product direction comes from the human. Execution should not block. When a call is the human's, apply a sensible default, record it in the artifact, and say what they could choose instead.