one-fix-one-pr-then-coordinatelisted

# One fix, one PR — then coordinate the cluster ## Purpose A real investigation rarely produces exactly one fix. You go in after one bug, read the subsystem, and come out with three — developed together, on one branch, touching the same files. The path of least resistance is to ship that branch: "here's my graceful-shutdown work." That forces a reviewer to evaluate unrelated changes as a unit, couples their merge and revert, and destroys the one thing a PR must answer — *what does this change do?* So the packaging is a separate act from the fixing: **split by issue, prove each stands alone, then coordinate.** Coordination is the half people skip. Three PRs appearing at once from a stranger reads as a dump; the same three, with the live PR answered first and one hub comment mapping the web, reads as someone who did a piece of work and made it navigable. The hub comment is the highest-value artifact in the whole cluster — it is the only place a maintainer can see the shape without reconstructing it. ## When to use - One thread of work yielded fixes for two or more filed issues. - You developed several fixes on one branch and now have to decide what to open. - A cluster is already open and nothing ties it together for a maintainer. ## When NOT to use The fixes are genuinely interdependent — B doesn't work without A — then one PR is correct, and say so in the body. Also skip when it's one fix for one issue: there is no cluster to coordinate, and a hub comment for a single