Skip to content

lidge-jun/codexclaw / plugins/codexclaw/skills/dev-code-reviewer

cxc-dev-code-reviewer

Unverified

Use for code/PR/diff review and refactor audits. Triggers: review this, before merge, antipattern, 리뷰, 코드 리뷰, 머지 전에 확인.

Dev-Code-Reviewer — Code Review Guide

Read dev first for project-wide conventions.

Class, fast path, rule authority, proof, and safety: dev. Current/public evidence: dev routing.

dev-testing owns test adequacy and QA execution. dev-debugging owns RCA; dev-architecture owns coupling and boundary placement.

Review Posture (REVIEW-POSTURE-01)

Review as a skeptical, independent outsider. Executor claims, passing tests, AI summaries, and user-facing "done" prose are untrusted until you confirm them yourself — assume the work may have failed and look for the regression or false-confidence test that proves it. Inspect artifacts before believing them; a green run you did not read is not evidence.

Output Contract (REVIEW-OUTPUT-01)

Tool findings go first (Automated Pre-Scan item 3); then manual findings sorted Critical > High > Medium > Low > Style; then a dedicated blocking_issues block; verdict last. For dispatched plan-audit (PABCD A-gate) reviews the verdict is additionally machine-scannable: end the reply with a final line VERDICT: PASS, VERDICT: GO-WITH-FIXES (blockers=N), or VERDICT: FAIL (mapping: Approve -> PASS; Approve-with-suggestions -> GO-WITH-FIXES; Request-changes / Block -> FAIL). The dispatching agent's exit rule is AUDIT-LOOP-01 (cxc-pabcd §A): FAIL always triggers another round. Every finding carries a concrete trigger, impact, and path:line (FAMILY-CITE-01) — no finding on a hunch. Do not file pre-existing debt unless the patch worsened it. When a change introduces a value/type/message crossing a module boundary, trace the consumer side before declaring it correct, rather than reviewing the emitting hunk alone.

Regression & false-confidence tests (REVIEW-REGRESS-01)

Run a dedicated pass: what previously-working behavior can now break, and do the tests cover that surface? Flag deletion-only "fixes", tautological tests, tests that merely mirror the implementation, and scope-drift abstractions added beyond the request.

2. Quality Thresholds

Severity Definitions

SeverityDefinition
CriticalExploitable security flaw, data loss, or production outage
HighCorrectness or security defect affecting users
MediumBounded defect or material maintainability risk
LowMinor risk with limited impact
StyleConvention-only issue with no behavioral risk

Review Verdict

IndicatorVerdictAction
No high/critical issues✅ ApproveMerge
Only Medium/Low/Style issues🔧 Approve with suggestionsFix Medium before merge unless the author explicitly marks it non-blocking with a stated reason and tracked follow-up
Any unresolved High issue⚠️ Request changesAuthor must address before merge
Any Critical issue🚫 BlockCannot merge until resolved

Deterministic blocker semantics (REVIEW-BLOCK-01): any unresolved Critical or High blocks the merge. Medium findings should be fixed before merge. When explicitly marked non-blocking by the author with a stated reason, Medium may pass with a tracked follow-up. Style never affects the verdict.

3. Common Antipatterns

Before writing or reviewing code smells, read common antipatterns.

Modular References

ConditionReference
Any review: pre-scan, impact order, mindset, readinessreview-process.md
Classify quality signals or file-size findingsquality-thresholds.md
Code-smell scan (dev §6)common-antipatterns.md
Every PR: security and performance quick-checkssecurity-performance-review.md
Receive/request feedback or review AI-generated codereview-feedback.md
Subagent review, AI evidence, slop cleanup, or boundary-guard deletionsubagent-and-slop-review.md
Before verdict; falsify findings, cover changed files, or re-review an interdifffinding-verification.md
Current/public review proof or browser-source verificationexternal-review-evidence.md
Tech debt inventory/paydowntech-debt.md
AI review tool coordinationai-assisted-review.md
rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit.Stars: 464kUpdated Claude CodeCodexVerifiedInstall
rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure.Stars: 464kUpdated Claude CodeCodexVerifiedInstall
rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics.Stars: 464kUpdated Claude CodeCodexVerifiedInstall
ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first.Stars: 192kUpdated Claude CodeCodexVerifiedInstall