Closed-loop execute (V-model right arm)

Mechanical verification pipeline. Every verify step emits evidence rows tied to acceptance criterion IDs (AC-n).

When to use

The harness is opt-in. Run it when:

• Invoked as /closed-loop <task> or /closed-loop <spec-path>.
• The user asks for the closed loop, proper verification, or an evidence trail.
• verification_harness: on in 00-inbox/MY-PROFILE.md and this is a build task.
• Another skill that declares a normal+ lane reaches its verify step.

Do not run it on a request that did not ask for it. Notes, briefs, research, drafts, and ordinary edits are not harness runs, and a checkpoint ledger on those is pure overhead.

Phase 0 — Lane + run folder

bash .claude/lib/lane-classify.sh explain "<task>"
bash .claude/lib/checkpoint.sh init 04-projects/harness/runs/<YYYY-MM-DD-HHmm>

Lane	Checkpoints
tiny	CP-3 → CP-5 (if mutating)
normal	CP-1 → CP-2 → CP-3 → CP-3v → CP-5
full	+ CP-4 + claim-verifier + CP-6
bug	root-cause ledger (CP-0) before CP-3

Record: checkpoint.sh record <run-dir> CP-0 PASS|SKIP "<lane>"

Phase 1 — CP-1 Spec (acceptance criteria)

If spec exists, use its ## Acceptance criteria + traceability matrix. Else write:

04-projects/harness/runs/<id>/criteria.md using references/spec-template.md (criteria + matrix sections only).

Each criterion: falsifiable + AC-n ID + verify method.

Record: checkpoint.sh record <run-dir> CP-1 PASS "N criteria"

Phase 2 — CP-2 Plan

Map tasks → AC IDs in evidence/CP-2-plan.md. Update matrix status to pending.

Record: checkpoint.sh record <run-dir> CP-2 PASS

Phase 3 — CP-3 Build

Worker implements. Returns deliverable path only.

Phase 4 — CP-3v Component verify

retry=0
loop:
  spawn task-verifier (fresh context, read-only)
  merge EVIDENCE rows into evidence/ledger.md
  if PASS → break
  if FAIL:escalate → record CP-3v FAIL, escalate
  if FAIL:fixable && retry < 2 → fix-agent → retry++
  else → escalate

Copy verifier EVIDENCE rows into evidence/CP-3v-component.md.

Record: checkpoint.sh record <run-dir> CP-3v PASS|FAIL

Phase 5 — CP-4 Integration verify (full or multi-task)

Spawn integration-verifier (read-only). Append rows to ledger.

Skip for single-task normal.

Record: checkpoint.sh record <run-dir> CP-4 PASS|SKIP

Phase 6 — CP-5 Acceptance (post-condition)

For each mutation, observe artifact (curl, screenshot, re-fetch). Emit:

EVIDENCE AC-n | CP-5 | PASS | <observation> | <artifact>

UI/UX flow changes: the post-condition is visual. Screenshot every meaningful state with whatever browser tooling the environment has, then read the image and confirm no overflow, misalignment, clipping, wrong color, or broken responsive layout before PASS. The Observation must describe what you saw; the artifact is the screenshot/GIF in evidence/. Fix any visual defect and re-capture. See CLAUDE.md → Visual Verification.

Write evidence/CP-5-acceptance.md. Traceability closure: every AC in matrix has ≥1 PASS row in ledger.

Record: checkpoint.sh record <run-dir> CP-5 PASS|FAIL

Phase 7 — Record + handoff

• Append to .claude/logs/loop-ledger.tsv
• Update spec traceability matrix statuses to verified
• full lane / big task: generate an HTML rollup from references/report-template.html → 04-projects/harness/runs/<id>/report.html, filled from criteria.md + evidence/ledger.md (criteria, AC traceability, verifier verdicts, post-condition observations). Self-contained; SendUserFile it or publish as an Artifact. Skip for normal/tiny.
• Suggest /retro <run-dir> for CP-7

Integration

Skill	Lane	CP-4
ultragoal	full per phase (never downgraded)	integration-verifier + north-star acceptance
team-brief	full	claim-verifier
comprehensive-analysis, auto-research	full	claim-verifier on cited claims
content-factory	normal	skip
review-cockpit	normal	skip; CP-6 is the user's approval per card

Escalation template

ESCALATED — <task>
Lane: <lane> | Last CP: <CP-n>
Evidence bundle: 04-projects/harness/runs/<id>/evidence/
Open AC IDs: <list without PASS rows>
Decision needed: <one question>