← ClaudeAtlas

principle-human-owns-the-endslisted

Apply when deciding whether to consult the user mid-run. The human decides what to build and what to ship; everything between runs autonomously and defers findings to the PR review.
bostonaholic/team · ★ 11 · AI & Automation · score 73
Install: claude install-skill bostonaholic/team
# The Human Owns the Ends The human's leverage is at the edges of the pipeline, not the middle. Two decisions stay human: what to build and what to ship. Everything between runs autonomously, and mid-run findings are never surfaced for approval — the PR review at the end is the checkpoint the system designates. **Why:** A human approving mechanics mid-flight spends judgment where it is cheap. Judgment is the scarce resource; move it to where it compounds and automate everything else. Every consult the middle avoids is judgment saved for the two decisions that were always the human's. **Pattern:** - Never pause an autonomous run to triage a finding: Blocking and Major work loops the fixer automatically; Minor-and-below lands in the PR body's review notes. Minor is not a wastebasket — it is the human's queue. - Never land on the system's own judgment: merging is always a human decision. Opening the PR is autonomous; landing it is not. - Ask at the ends, be autonomous in the middle: a question that can wait for PR review waits. A blocked run halts terminally and reports rather than asking permission to continue.