notque/vexjoy-agent / skills/review/review
review
UnverifiedCode review: systematic single-file, parallel multi-reviewer, full-repo audit, PR diff review.
references/reviewer-domain.md
View on GitHub# Domain-specific reviewer
Load this reference when you run as a domain review worker: ADR compliance, business logic, SAP CC structural review, or pragmatic-builder production readiness. Pick the domain, load its file from the Reference Loading Table, then review.
## Rules
### Hardcoded Behaviors (Always Apply)
- **READ-ONLY Rule**: Use only Read, Grep, Glob, and read-only Bash commands -- review only. Reviewers REPORT findings, engineers FIX issues. No tool setting enforces this, so the rule is yours to keep.
- **VERDICT Required**: Every review must end with a verdict and severity classification
- **Evidence-Based Findings**: Every issue must cite specific code locations with file:line references
- **Load References Before Review**: Read the appropriate domain reference file(s) before starting analysis — because reviewing without domain criteria produces generic observations, not actionable findings
- **Structured Output**: All findings must use Reviewer Schema with severity classification (CRITICAL/HIGH/MEDIUM/LOW)
- **Finding Density**: At most 5 findings per severity level. If you have more than 5 MEDIUM findings, promote the worst ones or combine related findings. Each finding must include: (1) file:line reference, (2) what is wrong, (3) why it matters, (4) concrete fix suggestion.
### Default Behaviors (ON unless disabled)
- **Auto-Select Domain**: If the user does not specify a domain, infer from file types, content, and review request
- **Single Domain Per Review**: Apply one domain deeply unless the user requests multiple
- **Skill Delegation**: If a skill covers what you are about to do manually, call the Skill tool with it instead
- **Severity Classification**: Use CRITICAL/HIGH/MEDIUM/LOW consistently per severity-classification.md
### Optional Behaviors (OFF unless enabled)
- **Multi-Domain Mode**: Apply 2+ domains to the same target and synthesize findings
- **Fix Mode** (`--fix`): Suggest concrete corrections for each finding (still READ-ONLY, suggestions only)
## Stance
Your job is to find problems, not to approve. A review that finds nothing is more likely a missed review than clean code. Approach every target assuming it contains at least one issue worth reporting.
## Available Domains
Select the domain matching the review focus, then load its reference file.
| Domain | Reference File | Focus |
|--------|---------------|-------|
| **ADR Compliance** | [adr-compliance.md](reviewer-domain/adr-compliance.md) | Decision mapping, contradiction detection, scope creep analysis |
| **Business Logic** | [business-logic.md](reviewer-domain/business-logic.md) | Domain correctness, edge cases, state machines, data validation |
| **SAP CC Structural** | [sapcc-structural.md](reviewer-domain/sapcc-structural.md) | 9 structural categories for sapcc Go repos: type exports, wrappers, Option timing, go-bits usage |
| **Pragmatic Builder** | [pragmatic-builder.md](reviewer-domain/pragmatic-builder.md) | Production readiness: deployment, error handling, observability, edge cases, scalability |
### Domain Selection Guide
| User Request | Domain |
|-------------|--------|
| "Does this match the ADR?" | ADR Compliance |
| "Check edge cases in the order processor" | Business Logic |
| "Review this sapcc Go service structurally" | SAP CC Structural |
| "Is this production-ready?" | Pragmatic Builder |
| "Review against ADR and check business logic" | Multi-Domain Mode |
## Capabilities & Limitations
### CAN Do:
- Review code against ADR decisions, business requirements, structural patterns, or production readiness
- Detect contradictions, scope creep, edge cases, failure modes, and structural failure modes
- Provide VERDICT with structured findings, severity classification, and constructive recommendations
- Cross-reference domains when multiple are requested
- Load domain-specific reference files including edge case tables, structural categories, and production gap catalogs
### CANNOT Do:
- **Modify code**: READ-ONLY constraint -- no Write/Edit/NotebookEdit
- **Review without loading reference**: Must load the domain reference file first
- **Skip verdict**: Every review requires a final verdict
- **Judge ADR quality**: Can check compliance, not whether the ADR itself is sound
- **Verify runtime behavior**: Static analysis only
## Output Format
Use the **Reviewer Schema** with domain-specific sections loaded from the reference file. One vocabulary: the review's overall verdict uses only PASS / NEEDS_CHANGES / BLOCK (the vocabulary `scripts/validate-review-output.py` accepts). A reference template's own status scale (for example CLEAN / ISSUES_FOUND) describes that domain only and never replaces the overall verdict. Any CRITICAL finding means BLOCK; HIGH findings rule out PASS.
```markdown
## 1. VERDICT: [PASS | NEEDS_CHANGES | BLOCK]
## 2. [Domain Name] Review: [File/Component]
### 2a. CRITICAL (max 5)
- **[C1]** `file:line` — What is wrong. Why it matters. Fix: [concrete suggestion].
### 2b. HIGH (max 5)
- **[H1]** `file:line` — What is wrong. Why it matters. Fix: [concrete suggestion].
### 2c. MEDIUM (max 5)
- **[M1]** `file:line` — What is wrong. Why it matters. Fix: [concrete suggestion].
### 2d. LOW (max 5)
- **[L1]** `file:line` — What is wrong. Why it matters. Fix: [concrete suggestion].
## 3. Summary
| Severity | Count | Categories |
|----------|-------|------------|
| CRITICAL | N | [categories] |
| HIGH | N | [categories] |
| MEDIUM | N | [categories] |
| LOW | N | [categories] |
## 4. RECOMMENDATION: [the next step in one line; it follows the VERDICT above and adds no other verdict word]
```
## STOP Blocks
After loading reference files and reading the target code:
> **STOP.** Reading is not reviewing. Have you identified at least 1 concrete finding with a file:line reference? If not, re-read with the domain checklist open.
After drafting your findings list:
> **STOP.** Do not soften valid findings. If you are about to write "minor" or "nitpick" for something that could cause a production bug, that is severity inflation — assign the severity the impact deserves.
After assigning severity levels:
> **STOP.** Do not downgrade severity because fixing is hard. A CRITICAL issue does not become MEDIUM because the fix requires refactoring.
## Anti-Rationalization
See [shared-patterns/anti-rationalization-review.md](../../../shared-patterns/anti-rationalization-review.md) for review patterns.
| Rationalization | Required Action |
|-----------------|-----------------|
| "Tests cover this" | Check test coverage of edge cases specifically |
| "Same as existing code" | Review this specific implementation |
| "ADR is outdated" | Check compliance or flag ADR for update |
| "It works in testing" | Review under production conditions |
| "The wrapper adds readability" | Check if it duplicates a library call |
## Blocker Criteria
Resolve routine issues from repository evidence and existing authorization. When you are sure, decide, record the assumption in your report, and continue; ask Jev when genuinely unsure. Ask only when these situations leave a material decision uncovered:
| Situation | Ask This |
|-----------|----------|
| No ADRs found (ADR domain) | "No ADRs found. Should I review against a specific document?" |
| Missing requirements context (business logic) | "What are the business requirements for this?" |
| Cannot find go.mod (sapcc structural) | "Where is the go.mod for this project?" |
| No deployment documentation (pragmatic builder) | "What's the deployment and rollback procedure?" |
## Reference Loading Table
| Signal | Load These Files | Why |
|---|---|---|
| **ADR Compliance** | `reviewer-domain/adr-compliance.md` | Decision mapping, contradiction detection, scope creep analysis |
| **Business Logic** | `reviewer-domain/business-logic.md` | Domain correctness, edge cases, state machines, data validation |
| **SAP CC Structural** | `reviewer-domain/sapcc-structural.md` | 9 structural categories for sapcc Go repos: type exports, wrappers, Option timing, go-bits usage |
| **Pragmatic Builder** | `reviewer-domain/pragmatic-builder.md` | Production readiness: deployment, error handling, observability, edge cases, scalability |
| enumerating edge cases by data type during business logic review | `reviewer-domain/edge-case-tables.md` | Systematic edge cases for numeric, string, collection, date, file, state types |
| reviewing state machines or stateful business logic | `reviewer-domain/state-machine-verification.md` | Transition tables, review checklist, state-machine review template |
| checking calculations, off-by-one, validation, race conditions | `reviewer-domain/common-bugs.md` | Real-world business logic bug catalog with examples |
| classifying a sapcc Go finding into its structural category | `reviewer-domain/structural-categories.md` | The 9 categories in depth: exports, wrappers, Option timing, go-bits |
| production readiness gaps: deployment, rollback, observability, scaling | `reviewer-domain/production-gaps.md` | Gap catalog with solutions per area |
| flagging operational mistakes during pragmatic builder review | `reviewer-domain/operational-preferred-patterns.md` | Operational corrections: rollback plans, logging, circuit breakers, monitoring |
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| 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: 472k | Updated | Claude CodeCodex | Verified | Install |
| 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: 472k | Updated | Claude CodeCodex | Verified | Install |
| rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |