claude-flow-novice 2.0.0 → 2.0.1

package/config/cfn-loop/instructions/enterprise-instructions.md

@@ -0,0 +1,506 @@
+# CFN Loop — Enterprise Mode
+
+**Use when:** Production-critical systems, compliance, large scale, many stakeholders.
+
+---
+
+## Thresholds
+
+| Gate (L3) | Consensus (L2) | Loop 0.5 Design | Max Iters (L3/L2) | Coverage | PO |
+|---|---|---|---|---|---|
+| ≥0.75 | ≥0.95 | ≥0.85 architect consensus | 15 / 15 | ≥90% + e2e | 4-person board (weighted) |
+
+---
+
+## Loop Flow
+
+```
+Loop 0.5 (Planning - Enterprise Only)
+  → 3 architects → design consensus ≥0.85
+  → ADRs, system diagrams, API specs
+  → Redis debate channel, SQLite design storage
+
+Loop 3 (Implementation)
+  → Follow Loop 0.5 approved design
+  → Gate ≥0.75 each agent
+  → Max 15 iterations
+
+Loop 2 (Validation)
+  → 4 validators (quality, security, perf, tester)
+  → Consensus ≥0.95 (highest bar)
+  → Max 15 iterations
+
+Loop 4 (Product Owner Board - Enterprise Only)
+  → 4-person weighted board (CTO 30%, PO 30%, User 20%, A11y 20%)
+  → PROCEED / DEFER / ESCALATE
+  → Weighted voting ≥0.75
+```
+
+---
+
+## Loop 0.5: Planning Consensus (Enterprise-Only)
+
+**Purpose:** Architect team debates design BEFORE Loop 3 implementation.
+
+**Why critical:** Prevents costly rework via upfront design consensus.
+
+### Spawn Architects (Single Message)
+
+```bash
+Task: Spawn architect team for Loop 0.5 design consensus
+
+Agents:
+1. system-architect-persona (.claude/agents/planning-team/system-architect.yaml)
+   - Instruction: "Design system architecture. Focus: scalability (10M+ users), microservices boundaries. Publish to redis://design:debate:auth. Vote on final design."
+
+2. security-architect-persona (.claude/agents/planning-team/security-architect.yaml)
+   - Instruction: "Review design for threats. Focus: OAuth 2.0/OIDC, token security, threat modeling. Challenge insecure patterns on redis://design:debate:auth. Vote."
+
+3. api-designer-persona (.claude/agents/planning-team/api-designer.yaml)
+   - Instruction: "Design REST API contracts. Focus: versioning, OpenAPI 3.0, rate limiting. Propose on redis://design:debate:auth. Vote."
+
+All agents:
+- Subscribe: redis://design:debate:auth
+- Debate: 10-15 minutes
+- Voting: ≥0.85 consensus (33.3% each)
+- Output: ADRs, diagrams, API specs
+- Store: SQLite key "cfn/phase-auth/loop0.5/design" (ACL 3, 365d)
+```
+
+### Design Debate Process
+
+**Redis channel:** `design:debate:{phaseId}`
+
+**Message types:**
+- **Proposal:** Initial design with rationale, tradeoffs
+- **Challenge:** Security/scalability concerns, severity, recommendations
+- **Refinement:** Updated design addressing concerns
+- **Vote Request:** Call for consensus vote
+
+**Debate timeline:**
+- Minutes 0-5: Proposals
+- Minutes 5-10: Challenges and debate
+- Minutes 10-12: Refinements
+- Minutes 12-15: Voting
+
+### Voting & Consensus
+
+```typescript
+// Vote structure
+{
+  agent: "system-architect-persona",
+  vote: "approve",  // approve | concerns | reject
+  confidence: 0.90,
+  weight: 0.333,
+  reasoning: "Design meets scalability requirements"
+}
+
+// Consensus calculation (≥0.85 to proceed)
+const consensus = (
+  architect1.confidence * 0.333 +
+  architect2.confidence * 0.333 +
+  architect3.confidence * 0.333
+);
+```
+
+**Example:**
+```
+System Architect: approve (0.90) × 0.333 = 0.300
+Security Architect: approve (0.85) × 0.333 = 0.283
+API Designer: concerns (0.80) × 0.333 = 0.266
+Consensus: 0.849 ❌ (below 0.85)
+
+Action: Spawn mediator agent
+```
+
+### Output Storage
+
+**SQLite** (ACL Level 3: Swarm, 365-day TTL):
+
+```bash
+/sqlite-memory store \
+  --key "cfn/phase-auth/loop0.5/design" \
+  --level swarm \
+  --data '{
+    "consensus":0.88,
+    "design":{
+      "architecture":"microservices",
+      "security":"OAuth 2.0 + OIDC, encrypted Redis",
+      "scalability":"10M+ users, 99.99% SLA"
+    },
+    "adrs":["ADR-001-microservices.md","ADR-002-oauth2.md"],
+    "votes":[
+      {"agent":"system-architect","confidence":0.90},
+      {"agent":"security-architect","confidence":0.88},
+      {"agent":"api-designer","confidence":0.85}
+    ]
+  }' \
+  --ttl 31536000
+```
+
+### Mediator (If Consensus <0.85)
+
+```bash
+Task: Spawn design mediator
+
+Agent: design-mediator-persona
+Instruction: "Consensus 0.849 below 0.85. API Designer concerns: backward compatibility. Facilitate negotiation. Propose: versioned API with 6-month deprecation. Target ≥0.85. Max 3 rounds."
+```
+
+---
+
+## Loop 3: Implementation (Following Design)
+
+**Must follow Loop 0.5 approved design** (stored in SQLite).
+
+### Spawn Implementers
+
+```bash
+Task: Spawn implementers for Loop 3
+
+Design Spec: Read from SQLite key "cfn/phase-auth/loop0.5/design"
+
+Constraints:
+- MUST follow microservices architecture (ADR-001)
+- MUST implement OAuth 2.0 + OIDC (ADR-002)
+- MUST use encrypted Redis
+- MUST achieve 90%+ test coverage
+
+Agents (mesh, 2-7):
+1. coder-backend-1: Implement auth-service per ADR-001
+   - OAuth 2.0 token issuance, encrypted Redis
+   - Store: SQLite "cfn/phase-auth/loop3/agent-1"
+   - Confidence: ≥0.75
+
+2. coder-backend-2: Implement token-service
+   - JWT signing, token refresh logic
+   - Store: SQLite "cfn/phase-auth/loop3/agent-2"
+   - Confidence: ≥0.75
+
+3. security-specialist: Validate security vs threat model
+   - Store: SQLite "cfn/phase-auth/loop3/agent-3"
+   - Confidence: ≥0.75
+
+4. tester: Integration tests, 90%+ coverage
+   - Store: SQLite "cfn/phase-auth/loop3/agent-4"
+   - Confidence: ≥0.75
+
+All: Coordinate via redis://swarm:coordination:auth
+```
+
+**Hierarchical** (if >7 agents): Coordinator + teams (up to 50 agents)
+
+---
+
+## Loop 2: 4-Validator Consensus (≥0.95)
+
+**After Loop 3 gate passes, spawn 4 validators:**
+
+```bash
+Task: Spawn 4-validator team
+
+Validators:
+1. code-quality-validator: SOLID, design patterns, maintainability
+   - Weight: 25%
+   - Target: ≥0.90
+
+2. security-specialist: OAuth review, threat model verification, pen test
+   - Weight: 25%
+   - Target: ≥0.90
+
+3. perf-analyzer: Load testing (10M+ users), Redis latency, scaling
+   - Weight: 25%
+   - Target: ≥0.90
+
+4. tester: 90%+ coverage, integration tests, edge cases
+   - Weight: 25%
+   - Target: ≥0.90
+
+All: Read design (Loop 0.5) + implementation (Loop 3)
+Consensus target: ≥0.95 (weighted average)
+```
+
+### Consensus Calculation
+
+```typescript
+const consensus = (
+  0.92 * 0.25 +  // code-quality
+  0.95 * 0.25 +  // security
+  0.94 * 0.25 +  // performance
+  0.96 * 0.25    // testing
+) = 0.9425
+
+// ALWAYS proceed to Loop 4 regardless of consensus
+// Product Owner Board can override validator recommendations
+// Validators provide recommendations, not gate-blocking decisions
+```
+
+**Result:** 0.9425 < 0.95 → Proceed to Loop 4 Board (Board can override)
+
+**Redis event:**
+
+```bash
+# Publish consensus result (always proceed to Loop 4)
+redis-cli publish "cfn:loop2:consensus:complete" '{
+  "consensus":0.9425,
+  "target":0.95,
+  "met":false,
+  "validators":[
+    {"name":"code-quality","confidence":0.92},
+    {"name":"security","confidence":0.95},
+    {"name":"perf","confidence":0.94},
+    {"name":"tester","confidence":0.96}
+  ],
+  "recommendations":[
+    {"severity":"low","description":"Refactor OAuth config"},
+    {"severity":"medium","description":"Add rate limiting"}
+  ],
+  "action":"proceed_to_loop4_board"
+}'
+```
+
+---
+
+## Loop 4: Multi-Stakeholder Board (Enterprise-Only)
+
+**Purpose:** 4-person weighted voting board makes decision.
+
+**Why critical:** Enterprise requires multi-stakeholder alignment.
+
+### Board Composition
+
+| Stakeholder | Weight | Primary Concerns |
+|-------------|--------|------------------|
+| CTO | 30% | Technical debt, scalability, security, cost |
+| Product Owner | 30% | Feature completeness, time-to-market |
+| Power User | 20% | UX, performance, workflows |
+| Accessibility Advocate | 20% | WCAG AAA, screen readers, keyboard nav |
+
+### Spawn Board
+
+```bash
+Task: Spawn 4-person Product Owner Board
+
+Context:
+- Loop 2 consensus: 0.9425 (below 0.95 by 0.0075)
+- Recommendations:
+  [LOW] Refactor OAuth config
+  [MEDIUM] Add rate limiting (10 req/min)
+
+Board:
+1. cto-agent (30%): Evaluate if 0.0075 gap justifies delay
+   - Vote: PROCEED/DEFER/ESCALATE
+   - Store: SQLite "cfn/phase-auth/loop4/vote-cto"
+
+2. product-owner-agent (30%): Time-to-market vs perfection
+   - Store: SQLite "cfn/phase-auth/loop4/vote-po"
+
+3. power-user-persona (20%): Rate limiting UX impact
+   - Store: SQLite "cfn/phase-auth/loop4/vote-user"
+
+4. accessibility-advocate-persona (20%): WCAG AAA compliance
+   - Store: SQLite "cfn/phase-auth/loop4/vote-a11y"
+
+All: Read all loop memory from SQLite
+Decision threshold: ≥0.75 weighted consensus
+```
+
+### Weighted Vote Calculation
+
+```typescript
+// Vote mapping: PROCEED=1.0, DEFER=0.5, ESCALATE=0.0
+const weightedScore = (
+  0.5 * 0.30 * 0.80 +  // CTO: DEFER × 30% × confidence
+  1.0 * 0.30 * 0.85 +  // PO: PROCEED × 30%
+  0.5 * 0.20 * 0.75 +  // User: DEFER × 20%
+  1.0 * 0.20 * 0.90    // A11y: PROCEED × 20%
+) = 0.630
+
+// Decision ranges
+if (weightedScore >= 0.75) decision = "PROCEED";
+else if (weightedScore >= 0.40) decision = "DEFER";
+else decision = "ESCALATE";
+```
+
+**Result:** 0.630 → **DEFER** (Add rate limiting, backlog refactor, launch agents for next phase)
+
+**Board can override validators:** Even if consensus <0.95, Board can vote DEFER if validator concerns are out-of-scope, enhancements, or can be backlogged.
+
+**Redis events:**
+
+```bash
+# Publish Board decision
+redis-cli publish "cfn:loop4:board:decision" '{
+  "decision":"DEFER",
+  "weighted_score":0.630,
+  "consensus":0.9425,
+  "override":true,
+  "reasoning":"Low-priority refactor out-of-scope, defer rate limiting to post-launch",
+  "action":"launch_next_phase_agents",
+  "votes":[
+    {"stakeholder":"cto","vote":"DEFER","confidence":0.80},
+    {"stakeholder":"po","vote":"PROCEED","confidence":0.85},
+    {"stakeholder":"user","vote":"DEFER","confidence":0.75},
+    {"stakeholder":"a11y","vote":"PROCEED","confidence":0.90}
+  ]
+}'
+
+# Publish phase complete
+redis-cli publish "cfn:phase:complete" '{
+  "phase":"auth",
+  "decision":"DEFER",
+  "boardOverride":true,
+  "nextPhase":"user-profile",
+  "action":"spawn_agents"
+}'
+
+# Publish phase transition
+redis-cli publish "cfn:phase:transition" '{
+  "from":"auth",
+  "to":"user-profile",
+  "trigger":"board_defer_decision",
+  "consensusOverride":true
+}'
+```
+
+### Disagreement Handling
+
+**Trigger:** Disagreement >0.15 between any two stakeholders
+
+```bash
+# If triggered, spawn facilitator
+Task: Spawn board facilitator
+
+Agent: board-facilitator-persona
+Instruction: "Disagreement: CTO (DEFER 0.80) vs PO (PROCEED 0.85). Issue: Rate limiting priority. Facilitate: Can it be hotfix post-launch? Max 3 rounds. Target >0.85."
+```
+
+### Decision Storage
+
+**SQLite** (ACL Level 4: Project, 365-day TTL):
+
+```bash
+/sqlite-memory store \
+  --key "cfn/phase-auth/loop4/decision" \
+  --level project \
+  --data '{
+    "decision":"DEFER",
+    "weighted_score":0.630,
+    "consensus":0.9425,
+    "override":true,
+    "votes":[
+      {"stakeholder":"cto","vote":"DEFER","confidence":0.80,"weight":0.30},
+      {"stakeholder":"po","vote":"PROCEED","confidence":0.85,"weight":0.30},
+      {"stakeholder":"user","vote":"DEFER","confidence":0.75,"weight":0.20},
+      {"stakeholder":"a11y","vote":"PROCEED","confidence":0.90,"weight":0.20}
+    ],
+    "validator_recommendations":["Refactor OAuth config","Add rate limiting"],
+    "backlog":["Refactor OAuth config","Rate limiting v2"],
+    "nextAction":"launch agents for user-profile phase",
+    "audit_trail":"Board decision for compliance"
+  }' \
+  --ttl 31536000
+```
+
+---
+
+## Quality Bars (Enterprise)
+
+### Security
+- Threat model complete, all threats mitigated
+- Penetration testing simulation
+- No hardcoded secrets
+- Encryption at rest + in transit
+- OAuth/OIDC compliance
+- HSM key storage
+
+### Performance
+- Load testing (10M+ users)
+- Horizontal scaling verified
+- Redis latency <5ms p99
+- Rate limiting where needed
+
+### Accessibility
+- WCAG 2.1 Level AAA
+- Screen reader compatibility
+- Keyboard navigation
+- Color contrast validated
+
+### Testing
+- Coverage ≥90%
+- Unit + integration + e2e tests
+- Load tests under stress
+
+### Documentation
+- ADRs complete
+- System diagrams (C4, sequence)
+- API docs (OpenAPI 3.0)
+- Runbooks for operations
+
+---
+
+## Retry Templates
+
+**Loop 0.5 retry** (consensus <0.85): Spawn mediator, negotiate, max 3 rounds
+
+**Loop 3 retry** (confidence <0.75): Replace with specialists, add missing roles, max 15 iterations
+
+**Loop 3 retry** (Board decides PROCEED): If Board votes PROCEED (not DEFER), relaunch Loop 3 with targeted fixes
+
+```bash
+# Publish retry event
+redis-cli publish "cfn:loop3:retry" '{
+  "iteration":4,
+  "reason":"Board voted PROCEED - address rate limiting",
+  "boardDecision":"PROCEED",
+  "targetAgents":["perf-engineer"],
+  "blocking_concerns":["rate_limiting_required"]
+}'
+```
+
+**Loop 4 retry** (disagreement >0.15 unresolved): ESCALATE to human after 3 deliberation rounds
+
+---
+
+## Quick Reference
+
+**Thresholds:**
+- Gate: ≥0.75
+- Consensus: ≥0.95
+- Loop 0.5: ≥0.85
+- Loop 4 Board: ≥0.75 weighted
+- Coverage: ≥90%
+
+**Board Weights:**
+- CTO: 30%
+- PO: 30%
+- User: 20%
+- A11y: 20%
+
+**Memory Keys:**
+```
+Loop 0.5: cfn/phase-{id}/loop0.5/design (ACL 3, 365d)
+Loop 3: cfn/phase-{id}/loop3/agent-{id} (ACL 1, 30d)
+Loop 2: cfn/phase-{id}/loop2/consensus (ACL 3, 30d)
+Loop 4: cfn/phase-{id}/loop4/decision (ACL 4, 365d)
+```
+
+**Redis Channels:**
+```
+design:debate:{phaseId}            # Loop 0.5 architect debate
+cfn:phase:start                    # Phase begins
+cfn:phase:complete                 # Phase ends (after Board)
+cfn:phase:transition               # Phase → phase
+cfn:loop3:agent:lifecycle          # Agent spawn/terminate
+cfn:loop3:agent:complete           # Agent confidence
+cfn:loop3:gate:result              # Gate pass/fail
+cfn:loop2:validator:spawned        # Validator start
+cfn:loop2:validator:complete       # Validator result
+cfn:loop2:consensus:complete       # Consensus result (always → L4)
+cfn:loop4:board:decision           # Board decision
+cfn:loop3:retry                    # Retry iteration
+```
+
+---
+
+**Enterprise Philosophy:** Highest quality bar. Security, scalability, accessibility, compliance.