← ClaudeAtlas

evaluating-sprint-completelisted

Evaluate the eight-invariant sprint-complete predicate.
Acendas/shipyard · ★ 2 · AI & Automation · score 71
Install: claude install-skill Acendas/shipyard
# Evaluating Sprint Complete A sprint is complete only when eight invariants hold simultaneously. This skill is the gate `/ship-execute` runs after the last wave-completion verifier passes and before declaring the sprint shippable. It composes the wave-level guarantees into the sprint-level guarantee. ## When to Invoke `/ship-execute` Step 5 (Sprint Completion) calls this skill after: - The full-suite test command exits 0. - The sprint-boundary verify-probe has completed and its capture is on disk. - `/ship-review`'s Stage 0 multi-agent scanner pipeline has been invoked and converged (or escalated cleanly). Other entry points: - `/ship-review` may invoke this skill at the beginning of a sprint review as a fast pre-flight ("is this sprint actually done before I bother running the full review?"). - Manual invocation when a user wants to spot-check sprint completeness before declaring shippable. ## Inputs - `sprint_id` — e.g., `sprint-003` - `data_dir` — literal `<SHIPYARD_DATA>` path - `working_branch` — git branch - `sprint_base_sha` — working-branch HEAD when the sprint started (from SPRINT.md frontmatter `base_sha` or computed via `git merge-base`) - `sprint_head_sha` — current HEAD - `sprint_verify_capture` — path to the sprint-boundary verify-probe capture - `sprint_verify_exit_code` — exit code from sprint-boundary verify - `demo_probe_event_window_start` — ISO timestamp; only `acceptance_probe_completed` events at or after this count for Invariant 8 (typically SP