Skip to content

nWave-ai/nWave / nWave/skills/nw-par-review-criteria

nw-par-review-criteria

Verified

Quality dimensions and review checklist for devop reviews

DevOp Reviewer: Review Criteria

Apply criteria only to the authorized operational/design scope and actual deployment promise. This skill defines review dimensions, not a mandatory phase sequence or artifact set. Existing authority and explicit no-rollout applicability are valid. Project thresholds need their own evidence; percentages/time targets below are not universal gates. Use the role-owned platform assessment in nw-typesafe-system-one; local verified reasons and distinct real observations decide findings, and scalar scores never compensate blockers.

Load nw-cross-cutting-invariants and apply flow:findings-go-to-the-owner-directly when exchanging findings or questions between roles.

Critique Dimension 1: Handoff adequacy

Compare the supplied authority/evidence against the operational decisions actually consumed by this task. A missing load-bearing interface, environment or recovery fact is a concrete finding; no fixed DISCUSS/DESIGN/DISTILL/DELIVER artifact or approval set is mandatory. DESIGN self-check is not an independent approval. Severity follows the consequence of the specific gap, not a phase label.


Critique Dimension 2: Deployment Readiness Gaps

Pattern: Feature marked "ready" but missing production prerequisites.

Require only the selected rollout's prerequisites: relevant current tests, configuration, promised telemetry, recovery and operational responsibilities. A document-only/no-rollout plan may be complete while live readiness remains unobserved; never require production validation outside authorization.

Severity: critical. Complete missing prerequisite before marking deployment-ready.


Critique Dimension 3: Traceability Violations

Pattern: Cannot trace production code back to requirements.

Required: User stories map to acceptance tests | Acceptance tests map to production code | Code changes traceable to commits | All AC verified in production.

Severity: high. Establish traceability chain: user-story -> acceptance-tests -> code-commits.


Critique Dimension 4: Priority Validation

Purpose: Validate roadmap addresses largest bottleneck first, not secondary concern.

Questions

Q1: Is this the largest bottleneck? Does timing data show primary problem? Larger problem being ignored? Assessment: YES / NO / UNCLEAR.

Q2: Were simpler alternatives considered? Roadmap includes rejected alternatives? Rejection reasons evidence-based? Simpler solution achieves 80% benefit? Assessment: ADEQUATE / INADEQUATE / MISSING.

Q3: Is constraint prioritization correct? Constraints quantified by impact? Architecture addresses constraint-free opportunities first? Minority constraint dominating? (flag if >50% of solution for <30% of problem). Assessment: CORRECT / INVERTED / NOT_ANALYZED.

Q4: Is architecture data-justified? Key architectural decision supported by quantitative data? Different data leads to different architecture? Assessment: JUSTIFIED / UNJUSTIFIED / NO_DATA.

Failure Conditions

  • FAIL if Q1 = NO (wrong problem being addressed)
  • FAIL if Q2 = MISSING (no alternatives considered)
  • FAIL if Q3 = INVERTED (minority constraint dominating)
  • FAIL if Q4 = NO_DATA and this is performance optimization

Critique Dimension 5: Functional Integration

Purpose: Verify feature wired into system entry point -- prevents Testing Theatre. A feature with 100% test coverage but 0% wiring tests is not complete.

Validation Criteria:

  1. Wiring test exists: at least one acceptance test invokes feature through driving port
  2. Component integrated: implemented component called from entry point module
  3. Boundary correct: acceptance tests do not import internal components directly

Gate failure response: Block finalization | report specific integration gap with evidence | require integration step before completion.


Quality Gate Checklist

Applicable technical and architecture evidence

  • Declared obligations have native tests/observations through their real boundaries.
  • Required project checks and review are current; source/oracle/runtime identity is known.
  • Interfaces, invariant owners, security and build/discovery agree with the selected design.
  • Coverage, performance and security measures are supplied only where applicable, under project criteria; no universal coverage percentage or tooling requirement.

Applicable operational evidence

  • A selected live rollout has concrete promotion/failure/rollback conditions.
  • Recovery and persistent-data claims distinguish designed plans from observed runs.
  • Promised KPI/SLO signals trace to collection, alert/runbook and responsibility.
  • Training/support/runbooks are required only when the authorized scope needs them.

Scope and readiness

  • All original promises and authorized human exclusions/amendments are respected.
  • No producer-only reduction, missing required boundary or stale evidence is approved.
  • Operational document review authorizes no deploy, infrastructure mutation or successor.
  • Unknown/not-applicable carries a concrete cause; a no-rollout feature buys no fake IaC.
receiving-code-reviewobra/superpowersUse when receiving code review feedback, before implementing suggestions, especially if feedback seems unclear or technically questionable - requires technical rigor and verification, not performative agreement or blind implementationStars: 322kUpdated Claude CodeCodexVerifiedInstall
requesting-code-reviewobra/superpowersUse when completing tasks, implementing major features, or before merging to verify work meets requirementsStars: 322kUpdated Claude CodeCodexVerifiedInstall
using-git-worktreesobra/superpowersUse when starting feature work that needs isolation from current workspace or before executing implementation plans - ensures an isolated workspace exists via native tools or git worktree fallbackStars: 322kUpdated Claude CodeCodexVerifiedInstall
verification-before-completionobra/superpowersUse when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before assertions alwaysStars: 322kUpdated Claude CodeCodexVerifiedInstall