@namch/agent-assistant 1.1.0 → 1.2.0

package/commands/review/team.md

@@ -0,0 +1,353 @@
+---
+description: "🔺 Team Review — Golden Triangle adversarial collaboration for maximum quality code review"
+version: "2.0"
+category: quality
+execution-mode: execute
+---
+
+# /review:team — Golden Triangle Code Review
+
+> **MISSION**: Maximum quality code review through adversarial collaboration.
+> Each phase spawns a **Golden Triangle** of 3 agents: Tech Lead (coordinator),
+> Executor (implementer), Reviewer (devil's advocate). Findings are released ONLY
+> upon consensus after debate.
+
+<scope>$ARGUMENTS</scope>
+
+---
+
+## 🛑 PRE-FLIGHT (DO FIRST — BLOCKS PHASE 1)
+
+**LOAD now** (in order; path `./rules/` or `~/.{TOOL}/skills/agent-assistant/rules/`):
+
+1. CORE.md — Identity, Laws, Routing
+2. PHASES.md — Phase Execution
+3. AGENTS.md — Tiered Execution
+4. **TEAMS.md** — Golden Triangle protocol (MANDATORY)
+
+**⛔ Do not run Phase 1 until all are loaded.** Follow **all** rules in those files; they override any conflicting instructions in this file.
+
+**Skills Resolution**: Load `SKILLS.md` on-demand for fitness calculation and dynamic discovery.
+
+---
+
+## 🔀 TIERED EXECUTION
+
+> Reference: AGENTS.md (Tiered Execution) + TEAMS.md (Golden Triangle Protocol)
+
+| Tier       | When                          | Action                                                                    |
+| ---------- | ----------------------------- | ------------------------------------------------------------------------- |
+| **TIER 1** | runSubagent/Agent Tool EXISTS | Orchestrator spawns Tech Lead → Tech Lead spawns Executor + Reviewer      |
+| **TIER 2** | Tool MISSING or SYSTEM error  | EMBODY Tech Lead → EMBODY Executor → EMBODY Reviewer → EMBODY Tech Lead  |
+
+**❌ Anti-Lazy**: Never use TIER 2 when TIER 1 tool available.
+
+**TIER 2 Golden Triangle Embodiment** (per TEAMS.md):
+```
+1. EMBODY Tech Lead → decompose → Shared Task List → dispatch
+2. EMBODY Executor → execute → post SUBMISSION to Mailbox
+3. EMBODY Reviewer → review → post REVIEW to Mailbox
+4. IF FAIL → EMBODY Executor → fix/defend → EMBODY Reviewer → re-check (max 3 rounds)
+5. EMBODY Tech Lead → arbitrate if needed → post DECISION → synthesize output
+```
+
+---
+
+## 📬 MAILBOX & DELIVERABLES
+
+**Mailbox**: `./reports/MAILBOX-{date}.md` — All triangle agents READ/APPEND. Never overwrite.
+**Protocol**: See TEAMS.md § Mailbox Message Types (TASK_ASSIGNMENT, SUBMISSION, REVIEW, DEFENSE, RESUBMISSION, APPROVAL, ESCALATION, ARBITRATION, DECISION).
+
+| Phase / Team     | Output                                          |
+| ---------------- | ----------------------------------------------- |
+| Phase 1 (Scope)  | `./reports/scouts/SCOUT-{scope}.md`             |
+| Phase 2 (Review) | Review findings in Mailbox                      |
+| Phase 3 (Plan)   | `./reports/reviews/REVIEW-REPORT-{scope}.md`    |
+| Phase 4 (Summary)| `./reports/reviews/REVIEW-{scope}.md`           |
+
+All files in `./reports/` → English only.
+
+---
+
+## 🔗 PHASE DEPENDENCIES
+
+| Phase                             | Requires                        | Blocking    |
+| --------------------------------- | ------------------------------- | ----------- |
+| P1: Scope & Context               | User scope / PR / files         | No          |
+| P2: Code Quality Review           | Scout findings from P1          | **YES**     |
+| P3: Improvement Plan              | All review findings from P2     | **YES**     |
+| P4: Summary & Recommendations     | Improvement plan from P3        | **YES**     |
+
+**⛔ Blocking**: If input missing → STOP → Create it first → Resume
+
+---
+
+## ⛔ INCREMENTAL EXECUTION (MANDATORY)
+
+One phase at a time. Within each phase follow the Golden Triangle Loop (TEAMS.md § Golden Triangle Protocol):
+
+```
+1. Spawn Golden Triangle (Tech Lead + Executor + Reviewer)
+2. Tech Lead decomposes → publishes Shared Task List
+3. Executor executes → posts SUBMISSION to Mailbox
+4. Reviewer critiques → posts REVIEW to Mailbox
+5. Debate loop: fix/defend → re-review (max 3 rounds)
+6. Consensus → Tech Lead posts DECISION → Phase output released
+```
+
+**Consensus Stamp** (required per phase): `✅ CONSENSUS: {TechLead} ✓ | {Executor} ✓ | {Reviewer} ✓`
+
+**Plan Compliance**: If `./reports/plans/PLAN-{scope}.md` exists → verify code matches plan, flag deviations.
+
+---
+
+## 🎭 Phase 1: SCOPE & CONTEXT — 🔺 GOLDEN TRIANGLE
+
+| Role      | Agent                                | Mission                                                    |
+| --------- | ------------------------------------ | ---------------------------------------------------------- |
+| Tech Lead | `reviewer`                           | Decompose: review scope, files to review, review dimensions |
+| Executor  | `scouter`                            | Execute: map changed files, identify blast radius, trace dependencies |
+| Reviewer  | `tech-lead` (Devil's Advocate)       | Challenge: scope complete? Missing files? Architecture impact assessed? |
+
+**Triangle Loop**:
+1. `reviewer` decomposes scope into areas: changed files, dependency graph, architecture impact, plan compliance baseline
+2. `reviewer` posts TASK_ASSIGNMENT → dispatches to `scouter`
+3. `scouter` surveys codebase → posts SUBMISSION with findings per area:
+   - All changed files with line-level diff summary
+   - Call chain trace: callers and callees of changed functions
+   - Blast radius: downstream components affected by changes
+   - Recent churn: files with high change frequency (risk indicator)
+   - Architecture layer mapping: which layers are touched
+4. `tech-lead` reviews each SUBMISSION → posts REVIEW:
+   - Is the blast radius complete or are downstream effects missing?
+   - Are there untouched files that SHOULD have changed (consistency gap)?
+   - Is the architecture impact properly assessed?
+   - Are there hidden coupling risks not surfaced?
+5. If FAIL → `scouter` addresses gaps or defends → RESUBMISSION/DEFENSE
+6. Max 3 rounds → ESCALATION to `reviewer` if unresolved
+7. `reviewer` synthesizes approved findings into unified scope document
+
+**Deliverable**: `./reports/scouts/SCOUT-{scope}.md`
+**Exit Criteria**: Review scope mapped, blast radius assessed, dependencies traced, focus areas flagged
+**Consensus**: ✅ CONSENSUS: reviewer ✓ | scouter ✓ | tech-lead ✓
+
+---
+
+## 🎭 Phase 2: CODE QUALITY REVIEW — 🔺 GOLDEN TRIANGLE (CRITICAL)
+
+> **THIS IS THE MOST CRITICAL PHASE.** Every step is detailed. No shortcuts.
+
+| Role      | Agent                                                     | Mission                                                    |
+| --------- | --------------------------------------------------------- | ---------------------------------------------------------- |
+| Tech Lead | `reviewer`                                                | Coordinate: assign review dimensions, synthesize findings  |
+| Executor  | `reviewer` (self-implements deep review)                  | Execute: line-by-line code review across ALL dimensions    |
+| Reviewer  | `security-engineer` + `performance-engineer` (combined)   | Challenge: security vulnerabilities? Performance concerns? Code standards? |
+
+**Prerequisite**: **READ** `./reports/scouts/SCOUT-{scope}.md`
+
+### 5 REVIEW DIMENSIONS (ALL MANDATORY)
+
+```
+┌──────────────────────────────────────────────────────────────┐
+│ 1. CORRECTNESS — Logic sound? Edge cases handled? Tests?     │
+│ 2. SECURITY (OWASP) — Injection, auth, input validation?    │
+│ 3. PERFORMANCE — N+1 queries, memory leaks, complexity?     │
+│ 4. ARCHITECTURE COMPLIANCE — Patterns, layer boundaries?     │
+│ 5. CODE QUALITY — Naming, DRY, SOLID, readability?          │
+└──────────────────────────────────────────────────────────────┘
+```
+
+### GOLDEN TRIANGLE REVIEW LOOP (CRITICAL — Step by Step)
+
+```
+╔══════════════════════════════════════════════════════════════════════╗
+║  PHASE 2: CODE REVIEW LOOP — FOLLOW EXACTLY                         ║
+╠══════════════════════════════════════════════════════════════════════╣
+║                                                                      ║
+║  STEP 1: Tech Lead reads SCOUT-{scope}.md                           ║
+║  - Load scope and blast radius into context                          ║
+║  - Identify all files requiring review                               ║
+║  - Prioritize by risk (highest first)                                ║
+║                                                                      ║
+║  STEP 2: Tech Lead creates Shared Task List                          ║
+║  - Break review into per-file/module tasks                           ║
+║  - Format: [ID] [Status] [File/Module] [Dimensions] [Risk Level]    ║
+║  - Post as TASK_ASSIGNMENT to Mailbox                                ║
+║                                                                      ║
+║  STEP 3: FOR EACH TASK — Executor performs deep review               ║
+║  a. Read TASK_ASSIGNMENT from Mailbox                                ║
+║  b. Review code line-by-line across ALL 5 dimensions                 ║
+║  c. Post SUBMISSION to Mailbox:                                      ║
+║     │ Task: {ID} | File(s): {list} | Lines: {range}                 ║
+║     │ Findings:                                                      ║
+║     │ | # | Dim  | Sev  | File:Line | Issue      |                   ║
+║     │ |---|------|------|-----------|------------|                   ║
+║     │ | 1 | SEC  | CRIT | app.ts:42 | SQLi       |                   ║
+║     │ | 2 | PERF | WARN | db.ts:18  | N+1 query  |                   ║
+║     │ Positive notes: {what's done well}                             ║
+║     │ Plan compliance: {if PLAN exists}                              ║
+║                                                                      ║
+║  STEP 4: Reviewer challenges each SUBMISSION (combined lens)         ║
+║  d. Read SUBMISSION from Mailbox                                     ║
+║  e. Apply triple lens:                                               ║
+║     SECURITY: OWASP missed? Auth gaps? Injection? Secrets?          ║
+║     PERFORMANCE: Complexity? N+1? Memory leaks? Caching?            ║
+║     STANDARDS: Naming? Error handling? Concurrency? Observability?  ║
+║  f. Post REVIEW to Mailbox:                                          ║
+║     │ Task: {ID} | Status: PASS or FAIL                             ║
+║     │ Additional findings: {table}                                   ║
+║     │ Disputed executor findings: {list}                             ║
+║     │ Confirmed executor findings: {list}                            ║
+║     │ Required actions: {list}                                       ║
+║                                                                      ║
+║  STEP 5: IF FAIL — Debate Loop (max 3 rounds)                       ║
+║  g. Executor reads REVIEW findings                                   ║
+║  h. For EACH finding:                                                ║
+║     - VALID → Accept, upgrade severity                               ║
+║     - DISPUTED → DEFENSE with evidence                               ║
+║  i. Post RESUBMISSION/DEFENSE:                                       ║
+║     │ Round: {1/2/3} | Accepted: {list} | Defended: {list+evidence} ║
+║  j. Reviewer re-reviews → back to (d)                                ║
+║  k. Round 3 unresolved → ESCALATION → Tech Lead ARBITRATION         ║
+║                                                                      ║
+║  STEP 6: IF PASS → APPROVAL → mark ✅ → next task                   ║
+║                                                                      ║
+║  STEP 7: ALL tasks complete                                          ║
+║  - Tech Lead consolidates: merge findings, deduplicate, group by dim ║
+║  - Post DECISION:                                                    ║
+║     │ Phase: 2 | Tasks: {n}/{total} | CRIT={n} WARN={n} INFO={n}   ║
+║     │ Disputes: {n} | Arbitrations: {n}                             ║
+║     │ ✅ CONSENSUS: reviewer ✓ | reviewer(exec) ✓ | security+perf ✓║
+║  - Release consolidated findings table                               ║
+╚══════════════════════════════════════════════════════════════════════╝
+```
+
+**PLAN COMPLIANCE** (if PLAN exists):
+```
+FOR EACH phase in PLAN-{scope}.md:
+  - Verify code implements plan spec → "Plan Phase X → Code Y → ✅/⚠️/❌"
+```
+
+**Exit Criteria**: All files reviewed across 5 dimensions, findings severity-ranked, security + performance validated
+**Consensus**: ✅ CONSENSUS: reviewer ✓ | reviewer(exec) ✓ | security+perf ✓
+
+---
+
+## 🎭 Phase 3: IMPROVEMENT PLAN — 🔺 GOLDEN TRIANGLE
+
+| Role      | Agent                                | Mission                                                    |
+| --------- | ------------------------------------ | ---------------------------------------------------------- |
+| Tech Lead | `tech-lead`                          | Synthesize: prioritize findings, create improvement plan   |
+| Executor  | `planner`                            | Execute: write REVIEW-REPORT-{scope}.md with prioritized issues |
+| Reviewer  | `reviewer`                           | Challenge: priorities correct? Severity accurate? Missing improvements? |
+
+**Prerequisite**: All Phase 2 findings consolidated in Mailbox.
+
+**Triangle Loop**:
+1. `tech-lead` reads Phase 2 findings → decomposes into improvement areas:
+   - Critical issues (must fix before merge)
+   - Warnings (should fix, tech debt if deferred)
+   - Suggestions (opportunistic improvements)
+   - Plan deviations (if PLAN exists)
+2. `tech-lead` posts TASK_ASSIGNMENT → dispatches to `planner`
+3. `planner` writes structured improvement plan → posts SUBMISSION:
+   - Prioritized issue table with effort estimates
+   - Fix recommendations per issue (specific, actionable)
+   - Dependency order: which fixes must come first
+   - Risk of NOT fixing: business + technical impact per issue
+4. `reviewer` reviews improvement plan → posts REVIEW:
+   - Severity ratings accurate or inflated/deflated?
+   - Fix recommendations correct and complete?
+   - Phase 2 findings dropped or downgraded without justification?
+   - Priority ordering sound (critical before cosmetic)?
+5. Debate loop if FAIL → `planner` adjusts or defends → max 3 rounds
+6. `tech-lead` synthesizes final improvement plan
+
+**Deliverable**: `./reports/reviews/REVIEW-REPORT-{scope}.md`
+**Exit Criteria**: All findings prioritized, fix recommendations documented, effort estimated, nothing dropped
+**Consensus**: ✅ CONSENSUS: tech-lead ✓ | planner ✓ | reviewer ✓
+
+---
+
+## 🎭 Phase 4: SUMMARY & RECOMMENDATIONS — 🔺 GOLDEN TRIANGLE
+
+| Role      | Agent                                | Mission                                                    |
+| --------- | ------------------------------------ | ---------------------------------------------------------- |
+| Tech Lead | `tech-lead`                          | Final synthesis: orchestrate executive summary creation     |
+| Executor  | `reporter`                           | Execute: executive summary, risk assessment, recommended actions |
+| Reviewer  | `business-analyst`                   | Challenge: business impact? Stakeholder communication? Actionable? |
+
+**Prerequisite**: **READ** `./reports/reviews/REVIEW-REPORT-{scope}.md`
+
+**Triangle Loop**:
+1. `tech-lead` decomposes summary: verdict, risk matrix, actions, stakeholder brief
+2. `tech-lead` posts TASK_ASSIGNMENT → dispatches to `reporter`
+3. `reporter` writes final review report → posts SUBMISSION:
+   ```markdown
+   # Code Review: {scope}
+   ## Verdict: ✅ APPROVED / ⚠️ APPROVED WITH CONDITIONS / ❌ CHANGES REQUIRED
+   ## Executive Summary
+   {2-3 sentence overview of code quality, key risks, recommendation}
+   ## Risk Assessment
+   | Risk | Severity | Likelihood | Business Impact | Mitigation |
+   ## Critical Issues ({count}) — MUST fix
+   | # | Category | File:Line | Description | Fix | Effort |
+   ## Warnings ({count}) — SHOULD fix
+   | # | Category | File:Line | Description | Fix | Effort |
+   ## Suggestions ({count}) — COULD improve
+   | # | Category | File:Line | Description | Fix | Effort |
+   ## Security Summary
+   Vulnerabilities: {count by severity} | OWASP: {categories}
+   ## Performance Summary
+   Bottlenecks: {count} | Impact: {description}
+   ## Plan Compliance (if applicable)
+   Phases verified: {X}/{Y} | Deviations: {list or "None"}
+   ## Recommended Actions
+   1. {action} — Owner: {who} — Priority: {P0/P1/P2}
+   ```
+4. `business-analyst` reviews final report → posts REVIEW:
+   - Business impact clearly communicated?
+   - Non-technical stakeholder would understand verdict?
+   - Recommended actions concrete and assignable?
+   - Risk assessment calibrated (not alarmist or dismissive)?
+5. Debate loop if FAIL → `reporter` adjusts or defends → max 3 rounds
+6. `tech-lead` approves final report
+
+**Deliverable**: `./reports/reviews/REVIEW-{scope}.md`
+**Exit Criteria**: Executive summary complete, verdict issued, all findings actionable, stakeholder-ready
+**Consensus**: ✅ CONSENSUS: tech-lead ✓ | reporter ✓ | business-analyst ✓
+
+---
+
+## ✅ COMPLETION
+
+Present final review report with consensus stamps from ALL phases:
+
+```markdown
+# 🔺 Golden Triangle Review Report: {scope}
+
+## Phase Results
+| Phase | Triangle | Consensus | Rounds |
+|-------|----------|-----------|--------|
+| P1: Scope & Context | reviewer / scouter / tech-lead | ✅ | {n} |
+| P2: Code Quality Review | reviewer / reviewer / security+perf | ✅ | {n} |
+| P3: Improvement Plan | tech-lead / planner / reviewer | ✅ | {n} |
+| P4: Summary | tech-lead / reporter / business-analyst | ✅ | {n} |
+
+## Debate Summary
+- Total submissions: {count}
+- First-pass approvals: {count}
+- Debates triggered: {count}
+- Arbitrations needed: {count}
+
+## Mailbox Reference
+Full debate history: `./reports/MAILBOX-{date}.md`
+```
+
+**Verdict Routing**:
+
+1. ✅ **Approved** — Code ready to merge
+2. ⚠️ **Approved with conditions** — Minor fixes needed → `/fix:fast`
+3. ❌ **Changes required** — Critical issues → `/fix:hard` or `/fix:team`
+4. 🔒 **Security block** — Security vulnerabilities must be resolved → `/fix:hard`
+5. 🧪 **Test gaps** — Insufficient coverage → `/test:team`

package/commands/review.md

@@ -33,6 +33,9 @@ IF scope is small (PR, few files):
 IF scope is large (codebase, architecture):
   → Route to /review:hard
 
+IF scope is large AND maximum quality with team collaboration needed:
+  → Route to /review:team
+
 IF unsure:
   → Default to /review:fast
 ```
@@ -45,6 +48,7 @@ IF unsure:
 | -------------- | ---------------------- |
 | `/review:fast` | Quick PR/file review   |
 | `/review:hard` | Deep codebase analysis |
+| `/review:team` | Maximum quality with parallel agent team collaboration |
 
 ---
 
@@ -59,6 +63,7 @@ IF unsure:
 
 1. ⚡ **Fast** → `/review:fast` — Quick review
 2. ⚡⚡⚡ **Hard** → `/review:hard` — Deep analysis
+3. 👥 **Team** → `/review:team` — Full team collaboration (parallel agents, maximum quality)
 
 ⏳ Awaiting selection...
 ```