@namch/agent-assistant 1.1.1 → 1.2.1

package/code-assistants/codex-assistant/skills/agent-assistant-report-hard/SKILL.md

@@ -0,0 +1,18 @@
+---
+name: agent-assistant-report-hard
+description: "⚡ Detailed Report — Comprehensive project analysis and documentation. Use when user types /report:hard."
+---
+
+# /report:hard
+
+> Activates Agent Assistant `/report:hard` workflow.
+
+## Rules
+
+1. **Load first**: `~/.codex/skills/agent-assistant/rules/CORE.md` — follow all 10 Laws
+2. **Agents**: Spawn via Codex native multi-agent (`~/.codex/agents/`)
+3. **Skills**: Resolve from `~/.codex/skills/agent-assistant/matrix-skills/`
+
+## Workflow
+
+**Load and follow**: `~/.codex/skills/agent-assistant/commands/report/hard.md`

package/code-assistants/codex-assistant/skills/agent-assistant-report-hard/agents/openai.yaml

@@ -0,0 +1,4 @@
+display_name: "/report:hard"
+short_description: "⚡ Detailed Report — Comprehensive project analysis and documentation"
+brand_color: "#3B82F6"
+allow_implicit_invocation: true

package/code-assistants/codex-assistant/skills/agent-assistant-report-team/SKILL.md

@@ -0,0 +1,18 @@
+---
+name: agent-assistant-report-team
+description: "🔺 Team Report — Golden Triangle adversarial collaboration. Use when user types /report:team."
+---
+
+# /report:team
+
+> Activates Agent Assistant `/report:team` workflow.
+
+## Rules
+
+1. **Load first**: `~/.codex/skills/agent-assistant/rules/CORE.md` — follow all 10 Laws
+2. **Agents**: Spawn via Codex native multi-agent (`~/.codex/agents/`)
+3. **Skills**: Resolve from `~/.codex/skills/agent-assistant/matrix-skills/`
+
+## Workflow
+
+**Load and follow**: `~/.codex/skills/agent-assistant/commands/report/team.md`

package/code-assistants/codex-assistant/skills/agent-assistant-report-team/agents/openai.yaml

@@ -0,0 +1,4 @@
+display_name: "/report:team"
+short_description: "🔺 Team Report — Golden Triangle adversarial collaboration"
+brand_color: "#14B8A6"
+allow_implicit_invocation: true

package/code-assistants/codex-assistant/skills/agent-assistant-review/SKILL.md

@@ -0,0 +1,18 @@
+---
+name: agent-assistant-review
+description: "🔍 Review Router — Route to code review workflows. Use when user types /review."
+---
+
+# /review
+
+> Activates Agent Assistant `/review` workflow.
+
+## Rules
+
+1. **Load first**: `~/.codex/skills/agent-assistant/rules/CORE.md` — follow all 10 Laws
+2. **Agents**: Spawn via Codex native multi-agent (`~/.codex/agents/`)
+3. **Skills**: Resolve from `~/.codex/skills/agent-assistant/matrix-skills/`
+
+## Workflow
+
+**Load and follow**: `~/.codex/skills/agent-assistant/commands/review.md`

package/code-assistants/codex-assistant/skills/agent-assistant-review/agents/openai.yaml

@@ -0,0 +1,4 @@
+display_name: "/review"
+short_description: "🔍 Review Router — Route to code review workflows"
+brand_color: "#10B981"
+allow_implicit_invocation: true

package/code-assistants/codex-assistant/skills/agent-assistant-review-fast/SKILL.md

@@ -0,0 +1,18 @@
+---
+name: agent-assistant-review-fast
+description: "⚡ Quick Review — Fast PR/file review. Use when user types /review:fast."
+---
+
+# /review:fast
+
+> Activates Agent Assistant `/review:fast` workflow.
+
+## Rules
+
+1. **Load first**: `~/.codex/skills/agent-assistant/rules/CORE.md` — follow all 10 Laws
+2. **Agents**: Spawn via Codex native multi-agent (`~/.codex/agents/`)
+3. **Skills**: Resolve from `~/.codex/skills/agent-assistant/matrix-skills/`
+
+## Workflow
+
+**Load and follow**: `~/.codex/skills/agent-assistant/commands/review/fast.md`

package/code-assistants/codex-assistant/skills/agent-assistant-review-fast/agents/openai.yaml

@@ -0,0 +1,4 @@
+display_name: "/review:fast"
+short_description: "⚡ Quick Review — Fast PR/file review"
+brand_color: "#10B981"
+allow_implicit_invocation: true

package/code-assistants/codex-assistant/skills/agent-assistant-review-hard/SKILL.md

@@ -0,0 +1,18 @@
+---
+name: agent-assistant-review-hard
+description: "⚡⚡⚡ Full Review — Deep code analysis with architecture review. Use when user types /review:hard."
+---
+
+# /review:hard
+
+> Activates Agent Assistant `/review:hard` workflow.
+
+## Rules
+
+1. **Load first**: `~/.codex/skills/agent-assistant/rules/CORE.md` — follow all 10 Laws
+2. **Agents**: Spawn via Codex native multi-agent (`~/.codex/agents/`)
+3. **Skills**: Resolve from `~/.codex/skills/agent-assistant/matrix-skills/`
+
+## Workflow
+
+**Load and follow**: `~/.codex/skills/agent-assistant/commands/review/hard.md`

package/code-assistants/codex-assistant/skills/agent-assistant-review-hard/agents/openai.yaml

@@ -0,0 +1,4 @@
+display_name: "/review:hard"
+short_description: "⚡⚡⚡ Full Review — Deep code analysis with architecture review"
+brand_color: "#10B981"
+allow_implicit_invocation: true

package/code-assistants/codex-assistant/skills/agent-assistant-review-team/SKILL.md

@@ -0,0 +1,18 @@
+---
+name: agent-assistant-review-team
+description: "🔺 Team Review — Golden Triangle adversarial collaboration. Use when user types /review:team."
+---
+
+# /review:team
+
+> Activates Agent Assistant `/review:team` workflow.
+
+## Rules
+
+1. **Load first**: `~/.codex/skills/agent-assistant/rules/CORE.md` — follow all 10 Laws
+2. **Agents**: Spawn via Codex native multi-agent (`~/.codex/agents/`)
+3. **Skills**: Resolve from `~/.codex/skills/agent-assistant/matrix-skills/`
+
+## Workflow
+
+**Load and follow**: `~/.codex/skills/agent-assistant/commands/review/team.md`

package/code-assistants/codex-assistant/skills/agent-assistant-review-team/agents/openai.yaml

@@ -0,0 +1,4 @@
+display_name: "/review:team"
+short_description: "🔺 Team Review — Golden Triangle adversarial collaboration"
+brand_color: "#10B981"
+allow_implicit_invocation: true

package/code-assistants/codex-assistant/skills/agent-assistant-test/SKILL.md

@@ -0,0 +1,18 @@
+---
+name: agent-assistant-test
+description: "🧪 Test Router — Route to testing workflows. Use when user types /test."
+---
+
+# /test
+
+> Activates Agent Assistant `/test` workflow.
+
+## Rules
+
+1. **Load first**: `~/.codex/skills/agent-assistant/rules/CORE.md` — follow all 10 Laws
+2. **Agents**: Spawn via Codex native multi-agent (`~/.codex/agents/`)
+3. **Skills**: Resolve from `~/.codex/skills/agent-assistant/matrix-skills/`
+
+## Workflow
+
+**Load and follow**: `~/.codex/skills/agent-assistant/commands/test.md`

package/code-assistants/codex-assistant/skills/agent-assistant-test/agents/openai.yaml

@@ -0,0 +1,4 @@
+display_name: "/test"
+short_description: "🧪 Test Router — Route to testing workflows"
+brand_color: "#10B981"
+allow_implicit_invocation: true

package/code-assistants/codex-assistant/skills/agent-assistant-test-fast/SKILL.md

@@ -0,0 +1,18 @@
+---
+name: agent-assistant-test-fast
+description: "⚡ Quick Test — Fast testing for specific scope. Use when user types /test:fast."
+---
+
+# /test:fast
+
+> Activates Agent Assistant `/test:fast` workflow.
+
+## Rules
+
+1. **Load first**: `~/.codex/skills/agent-assistant/rules/CORE.md` — follow all 10 Laws
+2. **Agents**: Spawn via Codex native multi-agent (`~/.codex/agents/`)
+3. **Skills**: Resolve from `~/.codex/skills/agent-assistant/matrix-skills/`
+
+## Workflow
+
+**Load and follow**: `~/.codex/skills/agent-assistant/commands/test/fast.md`

package/code-assistants/codex-assistant/skills/agent-assistant-test-fast/agents/openai.yaml

@@ -0,0 +1,4 @@
+display_name: "/test:fast"
+short_description: "⚡ Quick Test — Fast testing for specific scope"
+brand_color: "#10B981"
+allow_implicit_invocation: true

package/code-assistants/codex-assistant/skills/agent-assistant-test-focus/SKILL.md

@@ -0,0 +1,18 @@
+---
+name: agent-assistant-test-focus
+description: "🔒 Focused Test — Comprehensive QA with enforced context optimization. Use when user types /test:focus."
+---
+
+# /test:focus
+
+> Activates Agent Assistant `/test:focus` workflow.
+
+## Rules
+
+1. **Load first**: `~/.codex/skills/agent-assistant/rules/CORE.md` — follow all 10 Laws
+2. **Agents**: Spawn via Codex native multi-agent (`~/.codex/agents/`)
+3. **Skills**: Resolve from `~/.codex/skills/agent-assistant/matrix-skills/`
+
+## Workflow
+
+**Load and follow**: `~/.codex/skills/agent-assistant/commands/test/focus.md`

package/code-assistants/codex-assistant/skills/agent-assistant-test-focus/agents/openai.yaml

@@ -0,0 +1,4 @@
+display_name: "/test:focus"
+short_description: "🔒 Focused Test — Comprehensive QA with enforced context optimization"
+brand_color: "#10B981"
+allow_implicit_invocation: true

package/code-assistants/codex-assistant/skills/agent-assistant-test-hard/SKILL.md

@@ -0,0 +1,18 @@
+---
+name: agent-assistant-test-hard
+description: "⚡⚡⚡ Full Test — Comprehensive QA with quality gates. Use when user types /test:hard."
+---
+
+# /test:hard
+
+> Activates Agent Assistant `/test:hard` workflow.
+
+## Rules
+
+1. **Load first**: `~/.codex/skills/agent-assistant/rules/CORE.md` — follow all 10 Laws
+2. **Agents**: Spawn via Codex native multi-agent (`~/.codex/agents/`)
+3. **Skills**: Resolve from `~/.codex/skills/agent-assistant/matrix-skills/`
+
+## Workflow
+
+**Load and follow**: `~/.codex/skills/agent-assistant/commands/test/hard.md`

package/code-assistants/codex-assistant/skills/agent-assistant-test-hard/agents/openai.yaml

@@ -0,0 +1,4 @@
+display_name: "/test:hard"
+short_description: "⚡⚡⚡ Full Test — Comprehensive QA with quality gates"
+brand_color: "#10B981"
+allow_implicit_invocation: true

package/code-assistants/codex-assistant/skills/agent-assistant-test-team/SKILL.md

@@ -0,0 +1,18 @@
+---
+name: agent-assistant-test-team
+description: "🔺 Team Test — Golden Triangle adversarial collaboration. Use when user types /test:team."
+---
+
+# /test:team
+
+> Activates Agent Assistant `/test:team` workflow.
+
+## Rules
+
+1. **Load first**: `~/.codex/skills/agent-assistant/rules/CORE.md` — follow all 10 Laws
+2. **Agents**: Spawn via Codex native multi-agent (`~/.codex/agents/`)
+3. **Skills**: Resolve from `~/.codex/skills/agent-assistant/matrix-skills/`
+
+## Workflow
+
+**Load and follow**: `~/.codex/skills/agent-assistant/commands/test/team.md`

package/code-assistants/codex-assistant/skills/agent-assistant-test-team/agents/openai.yaml

@@ -0,0 +1,4 @@
+display_name: "/test:team"
+short_description: "🔺 Team Test — Golden Triangle adversarial collaboration"
+brand_color: "#10B981"
+allow_implicit_invocation: true

package/code-assistants/copilot-assistant/agent-assistant.agent.md

@@ -3,31 +3,6 @@ name: Agent Assistant
 description: Central Orchestration Brain for Multi-Agent System. Delegates through Commands → Agents → Skills.
 argument-hint: Type command + task, e.g. /cook implement login
 tools: all
-handoffs:
-  - label: "🚀 Cook"
-    agent: "Agent Assistant"
-    prompt: "/cook:hard "
-  - label: "📋 Plan"
-    agent: "Agent Assistant"
-    prompt: "/plan:fast "
-  - label: "🛠 Fix"
-    agent: "Agent Assistant"
-    prompt: "/fix:hard "
-  - label: "🐛 Debug"
-    agent: "Agent Assistant"
-    prompt: "/debug "
-  - label: "🧪 Test"
-    agent: "Agent Assistant"
-    prompt: "/test "
-  - label: "📝 Code"
-    agent: "Agent Assistant"
-    prompt: "/code "
-  - label: "📚 Docs"
-    agent: "Agent Assistant"
-    prompt: "/docs:core "
-  - label: "🎨 Design"
-    agent: "Agent Assistant"
-    prompt: "/design "
 ---
 
 # ⚡ AGENT ASSISTANT — ORCHESTRATOR PROTOCOL

package/commands/brainstorm/team.md

@@ -0,0 +1,295 @@
+---
+description: "🔺 Team Brainstorm — Golden Triangle adversarial collaboration for maximum quality ideation"
+version: "2.0"
+category: ideation
+execution-mode: execute
+---
+
+# /brainstorm:team — Golden Triangle Creative Brainstorming
+
+> **MISSION**: Maximum quality ideation through adversarial collaboration.
+> Each phase spawns a **Golden Triangle** of 3 agents: Tech Lead (coordinator),
+> Executor (generator/analyst), Reviewer (devil's advocate). Ideas are released
+> ONLY upon consensus after debate.
+>
+> **KEY DIFFERENCE**: In brainstorming, the adversarial loop EXPANDS thinking
+> before CONVERGING. Phase 1 Reviewer pushes for MORE ideas. Phase 2 Reviewer
+> challenges evaluation fairness. Debate is generative here.
+
+<topic>$ARGUMENTS</topic>
+
+---
+
+## 🛑 PRE-FLIGHT (DO FIRST — BLOCKS PHASE 1)
+
+**LOAD now** (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.**
+
+---
+
+## 🔀 TIERED EXECUTION
+
+| 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.
+
+---
+
+## 📬 MAILBOX — Central Communication Hub
+
+**Location**: `./reports/MAILBOX-{date}.md` — Append only, never overwrite.
+
+| Type              | Sender    | Receiver  | Purpose                                        |
+| ----------------- | --------- | --------- | ---------------------------------------------- |
+| TASK_ASSIGNMENT   | Tech Lead | Executor  | Assign task with requirements and context       |
+| SUBMISSION        | Executor  | Reviewer  | Submit completed work for review                |
+| REVIEW            | Reviewer  | Executor  | PASS or FAIL with findings table                |
+| DEFENSE           | Executor  | Reviewer  | Defend work against FAIL findings               |
+| RESUBMISSION      | Executor  | Reviewer  | Resubmit after addressing findings              |
+| APPROVAL          | Reviewer  | Tech Lead | Confirm task passes review criteria             |
+| ESCALATION        | Any       | Tech Lead | Escalate unresolvable disagreement              |
+| ARBITRATION       | Tech Lead | All       | Binding resolution of dispute                   |
+| DECISION          | Tech Lead | All       | Final phase decision with consensus stamp       |
+
+---
+
+## 📁 DELIVERABLE FILES
+
+| Phase              | Output                                          |
+| ------------------ | ----------------------------------------------- |
+| P1: Diverge        | `./reports/brainstorms/IDEAS-{topic}.md`        |
+| P2: Converge       | `./reports/brainstorms/ANALYSIS-{topic}.md`     |
+| P3: Synthesize     | `./reports/brainstorms/BRAINSTORM-{topic}.md`   |
+| ALL Phases         | `./reports/MAILBOX-{date}.md`                   |
+
+---
+
+## 🔗 PHASE DEPENDENCIES
+
+| Phase                              | Requires                    | Blocking    |
+| ---------------------------------- | --------------------------- | ----------- |
+| P1: Divergent Thinking             | User request / topic        | No          |
+| P2: Convergent Analysis            | IDEAS-{topic}.md from P1    | **YES**     |
+| P3: Synthesis & Recommendation     | ANALYSIS-{topic}.md from P2 | **YES**     |
+
+**⛔ Blocking**: If input missing → STOP → Create it first → Resume
+
+---
+
+## ⛔ INCREMENTAL EXECUTION (MANDATORY)
+
+One phase at a time. Per phase: Spawn Triangle → Decompose → Execute → Review → Debate → Consensus → Release.
+
+**Consensus Stamp** (required per phase): `✅ CONSENSUS: {TechLead} ✓ | {Executor} ✓ | {Reviewer} ✓`
+
+---
+
+## 🎭 Phase 1: DIVERGENT THINKING — 🔺 GOLDEN TRIANGLE
+
+| Role      | Agent                          | Mission                                                                   |
+| --------- | ------------------------------ | ------------------------------------------------------------------------- |
+| Tech Lead | `brainstormer`                 | Decompose: brainstorming dimensions, seed ideas, define exploration space |
+| Executor  | `researcher` + `scouter`       | Execute: generate ideas, research prior art, explore possibilities        |
+| Reviewer  | `tech-lead` (Devil's Advocate) | Challenge: ideas feasible? Missing categories? Thinking too narrow?       |
+
+> **⚡ REVIEWER MINDSET**: GENERATIVE, not reductive. FAIL = "you stopped too soon"
+> or "you missed an entire category." Push for MORE ideas, not fewer.
+
+**Triangle Loop**:
+1. `brainstormer` defines exploration dimensions → Shared Task List:
+   - Dimension mapping (technical, UX, business, unconventional angles)
+   - Seed ideas per dimension to prime generation
+   - Constraint identification: boundaries of the exploration space
+2. `brainstormer` posts TASK_ASSIGNMENT → dispatches to `researcher` + `scouter`
+3. `researcher` + `scouter` execute per dimension → post SUBMISSION:
+   - `scouter`: codebase patterns, integration points, prior attempts
+   - `researcher`: industry prior art, analogous solutions, competitor analysis
+   - Both: ideas per dimension with rationale and feasibility signal
+4. `tech-lead` reviews → posts REVIEW:
+   - Entire categories unexplored? Constrained by current architecture?
+   - Missing 10x "crazy" approaches? Hidden assumptions?
+5. If FAIL → expand or defend → max 3 rounds → ESCALATION if unresolved
+6. `brainstormer` synthesizes approved explorations into unified idea catalog
+
+**Deliverable**: `./reports/brainstorms/IDEAS-{topic}.md`
+- Raw idea catalog (minimum 8-12 distinct ideas across dimensions)
+- Prior art references, codebase context, assumptions stated
+- Each idea tagged: possible / stretch / moonshot
+
+**Exit Criteria**:
+- [ ] No dimension gaps (Reviewer confirmed)
+- [ ] Prior art researched with sources
+- [ ] Minimum 8 distinct ideas across multiple dimensions
+- [ ] `IDEAS-{topic}.md` created
+
+**Consensus**: ✅ CONSENSUS: brainstormer ✓ | researcher+scouter ✓ | tech-lead ✓
+
+---
+
+## 🎭 Phase 2: CONVERGENT ANALYSIS — 🔺 GOLDEN TRIANGLE (CRITICAL)
+
+> **THIS IS THE MOST CRITICAL PHASE.** Every step is detailed. No shortcuts.
+> Raw ideas become ranked, evidence-backed candidates here.
+
+| Role      | Agent                                | Mission                                                                        |
+| --------- | ------------------------------------ | ------------------------------------------------------------------------------ |
+| Tech Lead | `brainstormer`                       | Coordinate: define evaluation criteria, filter and group ideas, manage ranking  |
+| Executor  | `researcher`                         | Execute: deep analysis of top candidates, pros/cons, feasibility scoring       |
+| Reviewer  | `reviewer` + `business-analyst` lens | Challenge: evaluation fair? Bias in ranking? Hidden risks? Missing trade-offs? |
+
+**Prerequisite**: **READ** `./reports/brainstorms/IDEAS-{topic}.md`
+
+### GOLDEN TRIANGLE CONVERGENT ANALYSIS LOOP (Step by Step)
+
+```
+╔══════════════════════════════════════════════════════════════════════╗
+║  PHASE 2: CONVERGENT ANALYSIS — FOLLOW EXACTLY                      ║
+╠══════════════════════════════════════════════════════════════════════╣
+║                                                                      ║
+║  STEP 1: Tech Lead reads IDEAS-{topic}.md                            ║
+║  - Load full idea catalog, group by theme, define criteria + weights ║
+║                                                                      ║
+║  STEP 2: Tech Lead creates Shared Task List                          ║
+║  - Task per idea cluster: deep-dive analysis                         ║
+║  - Task: cross-cutting trade-off comparison                          ║
+║  - Task: feasibility scoring matrix                                  ║
+║  - Post to Mailbox as TASK_ASSIGNMENT                                ║
+║                                                                      ║
+║  STEP 3: Dispatch — each assignment includes:                        ║
+║  • Idea cluster to analyze + evaluation criteria + weights           ║
+║  • Required evidence depth + comparison requirements                 ║
+║                                                                      ║
+║  STEP 4: FOR EACH TASK — Executor analyzes                           ║
+║  a. Read TASK_ASSIGNMENT from Mailbox                                ║
+║  b. Deep analysis: pros/cons, feasibility, impact, dependencies      ║
+║  c. Post SUBMISSION:                                                 ║
+║     ┌──────────────────────────────────────┐                         ║
+║     │ Idea cluster | Feasibility {1-10}     │                         ║
+║     │ Impact {1-10} | Effort {S/M/L/XL}     │                         ║
+║     │ Pros + Cons (with evidence)           │                         ║
+║     │ Dependencies | Risks | Open questions │                         ║
+║     └──────────────────────────────────────┘                         ║
+║                                                                      ║
+║  STEP 5: Reviewer checks 5 dimensions per SUBMISSION                 ║
+║     ┌──────────────────────────────────────────────────────┐         ║
+║     │ 1. FAIRNESS — Criteria applied consistently?          │         ║
+║     │ 2. EVIDENCE — Claims backed by data/research?         │         ║
+║     │ 3. BIAS — Anchoring? Confirmation? Sunk cost?         │         ║
+║     │ 4. TRADE-OFFS — Hidden costs? Missed downsides?       │         ║
+║     │ 5. RISKS — Tail risks? Second-order effects?          │         ║
+║     └──────────────────────────────────────────────────────┘         ║
+║  Posts REVIEW: PASS or FAIL with findings table + required actions    ║
+║                                                                      ║
+║  STEP 6: IF FAIL — Debate Loop (max 3 rounds)                       ║
+║  g. Executor reads findings                                          ║
+║  h. For each: Revise (if valid) or DEFENSE with evidence             ║
+║  i. Post RESUBMISSION/DEFENSE → Reviewer re-reviews                  ║
+║  j. Round 3 without resolution → ESCALATION → Tech Lead ARBITRATES   ║
+║                                                                      ║
+║  STEP 7: IF PASS → APPROVAL → mark ✅ → next task                   ║
+║                                                                      ║
+║  STEP 8: ALL tasks complete                                          ║
+║  o. Tech Lead verifies cross-cluster coherence:                      ║
+║     - No grading drift, consistent comparisons, defensible rankings  ║
+║  p. Posts DECISION with consensus stamp → Phase output released      ║
+╚══════════════════════════════════════════════════════════════════════╝
+```
+
+**Deliverable**: `./reports/brainstorms/ANALYSIS-{topic}.md`
+- Evaluation criteria and weights
+- Per-idea deep analysis (feasibility, impact, effort, risks)
+- Cross-idea comparison matrix
+- Ranked shortlist of top 3-5 candidates
+- Debate record summary
+
+**Exit Criteria**:
+- [ ] All ideas evaluated against defined criteria
+- [ ] Evaluation fairness validated (no bias detected or addressed)
+- [ ] Each top candidate has evidence-backed pros/cons
+- [ ] Trade-offs documented — no hidden costs
+- [ ] Rankings defensible — Reviewer confirmed
+- [ ] `ANALYSIS-{topic}.md` created
+
+**Consensus**: ✅ CONSENSUS: brainstormer ✓ | researcher ✓ | reviewer ✓
+
+---
+
+## 🎭 Phase 3: SYNTHESIS & RECOMMENDATION — 🔺 GOLDEN TRIANGLE
+
+| Role      | Agent          | Mission                                                                    |
+| --------- | -------------- | -------------------------------------------------------------------------- |
+| Tech Lead | `brainstormer` | Synthesize: compile final recommendation from analysis                     |
+| Executor  | `reporter`     | Execute: write BRAINSTORM-{topic}.md with ideas, analysis, recommendations |
+| Reviewer  | `tech-lead`    | Challenge: recommendation sound? Options fairly presented? Decision-ready? |
+
+**Prerequisite**: **READ** `./reports/brainstorms/ANALYSIS-{topic}.md`
+
+**Triangle Loop**:
+1. `brainstormer` reads analysis → decomposes synthesis → Shared Task List:
+   - Executive summary with top recommendation
+   - Full idea catalog (all explored, not just winners)
+   - Analysis summary per top candidate + comparison matrix
+   - Risk assessment and recommended next steps
+2. `brainstormer` posts TASK_ASSIGNMENT → dispatches to `reporter`
+3. `reporter` writes each section → posts SUBMISSION:
+   - Decision-ready prose, not raw data dumps
+   - Recommendation with rationale tracing to evidence
+   - Alternatives presented fairly (not straw-manned)
+   - Actionable next steps per option
+4. `tech-lead` reviews → posts REVIEW:
+   - Recommendation supported by analysis or cherry-picked?
+   - Alternatives fairly presented? Decision-maker has full context?
+   - Risks clearly communicated, not buried?
+5. If FAIL → revise or defend → max 3 rounds → ESCALATION if unresolved
+6. `brainstormer` synthesizes approved sections into final document
+
+**Deliverable**: `./reports/brainstorms/BRAINSTORM-{topic}.md`
+- Executive summary and top recommendation
+- Complete idea catalog (explored / shortlisted / recommended)
+- Evidence-backed analysis with comparison matrix
+- Risk assessment and concrete next steps per option
+
+**Exit Criteria**:
+- [ ] Recommendation clearly stated with rationale
+- [ ] All options fairly presented (Reviewer confirmed)
+- [ ] Document is decision-ready — no ambiguity
+- [ ] Next steps actionable per option
+- [ ] `BRAINSTORM-{topic}.md` created
+
+**Consensus**: ✅ CONSENSUS: brainstormer ✓ | reporter ✓ | tech-lead ✓
+
+---
+
+## ✅ COMPLETION
+
+```markdown
+# 🔺 Golden Triangle Brainstorm Report: {topic}
+
+## Phase Results
+| Phase | Triangle | Consensus | Rounds |
+|-------|----------|-----------|--------|
+| P1: Divergent Thinking | brainstormer / researcher+scouter / tech-lead | ✅ | {n} |
+| P2: Convergent Analysis | brainstormer / researcher / reviewer | ✅ | {n} |
+| P3: Synthesis | brainstormer / reporter / tech-lead | ✅ | {n} |
+
+## Debate Summary
+- Total submissions: {count} | First-pass approvals: {count}
+- Debates triggered: {count} | Arbitrations: {count}
+- Ideas generated: {count} | Ideas shortlisted: {count}
+
+## Mailbox: `./reports/MAILBOX-{date}.md`
+
+## Next Actions
+1. ✅ **Done** — Brainstorm complete (triangle-validated)
+2. 📋 `/plan:team` → Plan implementation of top recommendation
+3. 🏗️ `/cook:team` → Build chosen approach
+4. 🔄 `/brainstorm:team` → Iterate with refined scope
+```

package/commands/brainstorm.md

@@ -33,6 +33,9 @@ IF topic is clear (quick ideas, simple exploration):
 IF topic is complex (research needed, deep analysis):
   → Route to /brainstorm:hard
 
+IF topic is complex AND maximum quality with team collaboration needed:
+  → Route to /brainstorm:team
+
 IF unsure:
   → Default to /brainstorm:fast
 ```
@@ -45,6 +48,7 @@ IF unsure:
 | ------------------ | -------------------------------------- |
 | `/brainstorm:fast` | Quick ideation, simple exploration     |
 | `/brainstorm:hard` | Research-backed ideation with analysis |
+| `/brainstorm:team` | Maximum quality with parallel agent team collaboration |
 
 ---
 
@@ -59,6 +63,7 @@ IF unsure:
 
 1. ⚡ **Fast** → `/brainstorm:fast` — Quick ideation
 2. ⚡⚡⚡ **Hard** → `/brainstorm:hard` — Deep exploration
+3. 👥 **Team** → `/brainstorm:team` — Full team collaboration (parallel agents, maximum quality)
 
 ⏳ Awaiting selection...
 ```