cadence: Skill for Claude Code

Phase build (subagent-driven, verification-first)

The core rule this pipeline exists to enforce: a subagent's "done, all tests passing" is not evidence. Subagents in this repo have twice claimed all-green over real failures. Every completion claim gets re-verified in the main thread before it is recorded.

Preconditions

• Loop position is BUILD (or DRAFT with an approved draft — run cadence progress).
• Read the DRAFT: .cadence/phases/<phase>/<id>-DRAFT.md. Note each task's files: boundary, verify: command, and done: AC-N mapping. Respect ## Boundaries (the Do-NOT list) absolutely.

Pipeline

1. Isolate. Enter a git worktree for the phase (native worktree tool preferred over manual git worktree add; convention: branch feat/<slug>, worktree under .claude/worktrees/<slug>). Before any commit in this pipeline, confirm git rev-parse --show-toplevel points at the worktree — a subagent has committed to the primary checkout by mistake. If this build is deliberately running parallel agents (this phase and another in flight at once), isolate to a worktree/branch from task 1 — before any subagent touches a file — not reactively after a files: conflict is noticed. Also confirm no other live session already holds this phase/draft (check for a running cadence process, an open terminal, or recent PROGRESS.json activity via cadence doctor) before dispatching into it; if one might still be running, resume that session in place instead of starting a second one against the same draft.
2. Plan waves. Run cadence dispatch plan --json. It computes dispatch waves from depends: edges and files:-overlap edges. Never parallelize two tasks whose files: overlap, even if the plan seems to allow it.
3. Per wave — implement. Dispatch one implementer subagent per task in the wave (parallel Task-tool calls). Each prompt must contain: the task's full DRAFT text, its files: boundary ("touch nothing else"), TDD order (failing test referencing the AC-N token inside an asserting it()/ test() block, then implementation), the exact verify: command, and an instruction to report the commands it ran with their real output.
4. Per task — review. Dispatch a fresh, independent reviewer subagent per completed task: adversarial, diff-scoped, checks boundary compliance, test honesty (assertions, not AC-N comment drops), and DRAFT conformance. Findings go back to a fix round, then re-review.
5. Per task — re-verify yourself. In the main thread: read the actual diff, then run the full pipeline yourself — pnpm turbo run lint typecheck test build (package-filtered while iterating; full before the wave closes). Only after this passes, record cadence done <T>. Never record DONE from a report.
6. Whole-branch review. After all waves: one fresh subagent reviews the entire branch diff against the DRAFT's Objective + ACs. It must return "ready to merge" with zero Critical/Important findings. History says this pass catches real defects the per-task reviews miss (an ordering bug, doc prose describing a pre-fix algorithm, stale code comments) — do not skip it for "simple" phases.
7. Settle, one commit. Run cadence settle run --auto — if it refuses, the gate is right until proven otherwise; fix the cause, don't reach for --force/--allow-*. Once it passes, stage everything together — source, tests, docs, the .changeset/*.md for this phase (feature PRs carry their own changeset, never deferred to the release PR), and the phase artifacts (-DRAFT.md, -PROGRESS.json, -SUMMARY.*, and -SUMMARY-snapshot.* if the phase produced any refused-attempt siblings — phase 247) — and make one commit (feat:/fix:, phase id in the subject). state.json/STATE.md stay gitignored, never committed. If the phase closes a Praxis recommendation, promote it to shipped in this same commit.
8. Land. Exit the worktree, then invoke the pr-land skill.