sprint_protocollisted

# sprint_protocol — what every Sprint participant does the same way Load this first in any Sprint turn, then your role skill (`sprint_prep`, `sprint_pln`, `sprint_dev`, `sprint_rev`). Use the simplest path the current durable state supports; treat authority, lifecycle preconditions, durable writes, and typed handoffs as hard boundaries and use judgment inside them. Repeat a read only when later activity could have changed it or the next command requires live revalidation. ## Lifecycle One Sprint binds one roadmap feature, exact governing spec revisions, a participant set (one Planner, Developers, Reviewers) each on one harness/model/effort route, and work units: editing lanes of spec tasks, each with one Developer and one Reviewer, ordered by dependencies and waves. `prepared` (editable) -> `armed` (ready lanes dispatch to Developers) <-> `paused` (relay off; restructure and reroute here) -> `completed` or `aborted` (terminal; nothing deleted). One Sprint is armed at a time. A lane: dispatched -> Developer builds, registers the PR, requests review -> Reviewer records a verdict -> Developer merges under the Sprint grant once `authorize-merge` returns live green + approved -> the merged handoff wakes the Planner, who dispatches what became ready. After the last lane the conformance Reviewer records the whole-Sprint report; the engine closes the Sprint, closes the Developer chats, and deletes the Sprint artifact directory. Planner and Reviewer chats persist; no worktree is re