---
name: review-operations
description: Review operations runbook for completeness and safety
summary: "Verifies the full deployment lifecycle is documented, monitoring covers latency/errors/saturation, alert thresholds have rationale, and common failure scenarios have runbook entries."
phase: "quality"
order: 940
dependencies: [operations]
outputs: [docs/reviews/review-operations.md, docs/reviews/operations/review-summary.md, docs/reviews/operations/codex-review.json, docs/reviews/operations/antigravity-review.json]
conditional: null
knowledge-base: [review-methodology, review-operations, multi-model-review-dispatch, review-step-template]
---

## Purpose
Review operations runbook targeting operations-specific failure modes: deployment
strategy gaps, missing rollback procedures, monitoring blind spots, unjustified
alerting thresholds, missing runbook scenarios, and DR coverage gaps.

At depth 4+, dispatches to external AI models (Codex, Antigravity) for
independent review validation.

## Inputs
- docs/operations-runbook.md (required) — runbook to review
- docs/system-architecture.md (required) — for deployment coverage

## Expected Outputs
- docs/reviews/review-operations.md — findings and resolution log
- docs/operations-runbook.md — updated with fixes
- docs/reviews/operations/review-summary.md (depth 4+) — multi-model review synthesis
- docs/reviews/operations/codex-review.json (depth 4+, if available) — raw Codex findings
- docs/reviews/operations/antigravity-review.json (depth 4+, if available) — raw Antigravity findings

## Quality Criteria
- (mvp) Deployment lifecycle fully documented (deploy, verify, rollback)
- (mvp) Monitoring verified against minimum set: latency, error rate, and saturation
- (deep) Alert thresholds have rationale
- (deep) Common failure scenarios have runbook entries
- (mvp) At least production environment operations documented
- (deep) Dev/staging/production environment differences documented in operations runbook
- (deep) Each health check endpoint specifies expected status code, response time SLA, failure thresholds
- (mvp) Every finding categorized P0-P3 (P0 = Breaks downstream work. P1 = Prevents quality milestone. P2 = Known tech debt. P3 = Polish.) with specific runbook section, metric, and issue
- (mvp) Fix plan documented for all P0/P1 findings; fixes applied to operations-runbook.md and re-validated
- (mvp) Downstream readiness confirmed — no unresolved P0 or P1 findings remain before security step proceeds
- (depth 4+) Multi-model findings synthesized: Consensus (all models agree), Majority (2+ models agree), or Divergent (models disagree — present to user for decision)

## Methodology Scaling
- **deep**: Full multi-pass review. Multi-model review dispatched to Codex and
  Antigravity if available, with graceful fallback to Claude-only enhanced review.
- **mvp**: Deployment coverage only.
- **custom:depth(1-5)**:
  - Depth 1: Monitoring and logging pass only (1 review pass)
  - Depth 2: Add deployment and rollback pass (2 review passes)
  - Depth 3: Add incident response and scaling passes (4 review passes)
  - Depth 4: Add external model review (4 review passes + external dispatch)
  - Depth 5: Multi-model review with reconciliation (4 review passes + multi-model synthesis)

## Mode Detection
Re-review mode if previous review exists. If multi-model review artifacts exist
under docs/reviews/operations/, preserve prior findings still valid.

## Update Mode Specifics

- **Detect**: `docs/reviews/review-operations.md` exists with tracking comment
- **Preserve**: Prior findings still valid, resolution decisions, multi-model review artifacts
- **Triggers**: Upstream artifact changed since last review (compare tracking comment dates)
- **Conflict resolution**: Previously resolved findings reappearing = regression; flag and re-evaluate