anchi-toolkit 1.0.0 → 1.2.0

package/docs/ORCHESTRATION_RUNTIME.md

@@ -1,173 +0,0 @@
-# Dynamic Orchestration Runtime
-
-> **Quy trình AI tự động điều phối agents/skills khi thực thi command**
-
----
-
-## 🔄 Execution Flow
-
-```
-┌─────────────────────────────────────────────────────────────┐
-│ USER INPUT: /do "Add payment with Stripe"                   │
-└─────────────────────────────────────────────────────────────┘
-                           ↓
-┌─────────────────────────────────────────────────────────────┐
-│ STEP 1: LOAD ORCHESTRATION CONFIG                           │
-│                                                             │
-│   Read: .cursor/agent-skill-index.yaml                      │
-│   Read: .cursor/orchestration.yaml                          │
-│   Read: .cursor/failure-modes.yaml                          │
-└─────────────────────────────────────────────────────────────┘
-                           ↓
-┌─────────────────────────────────────────────────────────────┐
-│ STEP 2: ANALYZE USER REQUEST                                │
-│                                                             │
-│   Keywords detected: ["payment", "stripe"]                  │
-│   Files affected: [*.ts, prisma/schema.prisma]              │
-│   Complexity: medium                                        │
-└─────────────────────────────────────────────────────────────┘
-                           ↓
-┌─────────────────────────────────────────────────────────────┐
-│ STEP 3: DYNAMIC AGENT SELECTION                             │
-│                                                             │
-│   Check command_mapping:                                     │
-│   ├── /do → default: fullstack-developer                   │
-│   └── conditionals: none matched                             │
-│                                                             │
-│   Check agent boundaries:                                    │
-│   └── fullstack-developer.boundaries:                        │
-│       └── NOT: security_audit → security-auditor            │
-│                                                             │
-│   Result:                                                   │
-│   ├── PRIMARY: fullstack-developer (confidence: 0.85)       │
-│   └── SECONDARY: security-auditor (payment = sensitive)     │
-└─────────────────────────────────────────────────────────────┘
-                           ↓
-┌─────────────────────────────────────────────────────────────┐
-│ STEP 4: DYNAMIC SKILL ACTIVATION                            │
-│                                                             │
-│   Scan skills.activation rules:                              │
-│                                                             │
-│   payment-integration:                                       │
-│   ├── keyword: ["payment", "stripe"] → MATCH ✓              │
-│   ├── package: ["stripe"] → CHECK package.json              │
-│   └── requires_security_review: true → FLAG                 │
-│                                                             │
-│   databases:                                                 │
-│   └── file: prisma/schema.prisma → MATCH ✓                  │
-│                                                             │
-│   better-auth:                                               │
-│   └── keyword: ["auth", "login"] → NO MATCH                 │
-│                                                             │
-│   Activated: [payment-integration, databases]               │
-└─────────────────────────────────────────────────────────────┘
-                           ↓
-┌─────────────────────────────────────────────────────────────┐
-│ STEP 5: CHECK FAILURE MODES                                 │
-│                                                             │
-│   low_confidence:                                            │
-│   └── confidence 0.85 > 0.5 → OK                            │
-│                                                             │
-│   security_concern:                                          │
-│   └── payment = sensitive → ESCALATE to security-auditor    │
-│                                                             │
-│   architecture_violation:                                    │
-│   └── Check preset contract → OK                            │
-└─────────────────────────────────────────────────────────────┘
-                           ↓
-┌─────────────────────────────────────────────────────────────┐
-│ STEP 6: DETERMINE RISK LEVEL                                │
-│                                                             │
-│   Check orchestration.yaml risk_levels:                      │
-│   ├── keywords: ["payment"] → HIGH risk                     │
-│   └── chain: security_chain                                  │
-│                                                             │
-│   Required:                                                 │
-│   ├── auto_approve: false                                    │
-│   ├── require_review: true                                   │
-│   └── require_lead_approval: true                            │
-└─────────────────────────────────────────────────────────────┘
-                           ↓
-┌─────────────────────────────────────────────────────────────┐
-│ STEP 7: BUILD EXECUTION CHAIN                               │
-│                                                             │
-│   Chain: security_chain                                      │
-│   [1] planner → Create plan                                 │
-│   [2] fullstack-developer → Implement                       │
-│   [3] security-auditor → Security review                    │
-│   [4] code-reviewer → Final review                          │
-└─────────────────────────────────────────────────────────────┘
-                           ↓
-┌─────────────────────────────────────────────────────────────┐
-│ STEP 8: LOG TO AUDIT                                         │
-│                                                             │
-│   Write to .ai-audit/YYYY-MM-DD.jsonl:                       │
-│   {                                                         │
-│     "command": "/do 'Add payment'",                         │
-│     "agents": ["fullstack-developer", "security-auditor"],  │
-│     "skills": ["payment-integration", "databases"],         │
-│     "risk_level": "high",                                    │
-│     "chain": "security_chain"                                │
-│   }                                                         │
-└─────────────────────────────────────────────────────────────┘
-                           ↓
-┌─────────────────────────────────────────────────────────────┐
-│ STEP 9: EXECUTE WITH ORCHESTRATION                          │
-│                                                             │
-│   Apply agents, skills, and chain to implementation         │
-└─────────────────────────────────────────────────────────────┘
-```
-
----
-
-## 📋 Quy Tắc Điều Phối
-
-### Rule 1: Agent Selection Priority
-```
-1. Check command_mapping.conditionals
-2. If no match → Use command_mapping.default
-3. Check agent.boundaries for escalation
-4. Check skill.requires_security_review
-```
-
-### Rule 2: Skill Activation
-```
-FOR EACH skill IN agent-skill-index.yaml:
-  IF skill.activation.keyword MATCHES request:
-    ACTIVATE skill
-  IF skill.activation.file_pattern MATCHES affected_files:
-    ACTIVATE skill
-  IF skill.activation.package EXISTS IN package.json:
-    ACTIVATE skill
-```
-
-### Rule 3: Risk Assessment
-```
-Scan request for risk_levels keywords:
-  low: ["typo", "rename", "format"]
-  medium: ["feature", "component", "api"]
-  high: ["payment", "auth", "admin"]
-  critical: ["delete", "drop", "production"]
-
-Apply corresponding chain and approvals.
-```
-
-### Rule 4: Failure Mode Check
-```
-BEFORE execution:
-  IF confidence < 0.5 → ASK user
-  IF security_concern → ESCALATE
-  IF architecture_violation → STOP
-  IF outside_expertise → ESCALATE to appropriate agent
-```
-
----
-
-## 🔗 Config Files
-
-| File | Purpose |
-|------|---------|
-| `.cursor/agent-skill-index.yaml` | Agent capabilities, skill activation |
-| `.cursor/orchestration.yaml` | Chains, risk levels, escalation |
-| `.cursor/failure-modes.yaml` | Stop conditions, restraints |
-| `.ai-audit/` | Audit logs |

package/docs/WALKTHROUGH.md

@@ -1,192 +0,0 @@
-# User Walkthrough: E-commerce Shop trong 5 Ngày
-
-> Mô phỏng hành trình user mới sử dụng anchi-toolkit từ đầu đến deploy
-
----
-
-## 📋 Profile
-
-| | |
-|---|---|
-| **User** | Minh (Junior Developer, 1 năm kinh nghiệm) |
-| **Goal** | Tạo website bán hàng online |
-| **Preset** | Rapid MVP |
-| **Platform** | Cursor AI |
-
----
-
-## 🗓️ Day 1: Setup & First Feature
-
-### 1.1 Cài Đặt (2 phút)
-
-```bash
-mkdir my-shop && cd my-shop
-npx anchi-toolkit install
-```
-
-Output:
-```
-✅ .cursor/commands/ - 18 commands
-✅ .cursor/agents/ - 19 agents
-✅ .cursor/skills/ - 35 skills
-✅ presets/ - 3 architecture presets
-```
-
-### 1.2 Khởi Tạo
-
-```
-/start
-→ Chọn 2 (Dự án mới)
-→ Chọn 1 (Rapid MVP)
-```
-
-AI tự động:
-- Tạo Next.js project
-- Setup Prisma + SQLite
-- Lưu context vào `.ai-memory/`
-
-### 1.3 Homepage
-
-```
-/do "Add homepage with hero section and featured products"
-```
-
-AI:
-- Phân tích → fullstack-developer (0.9)
-- Plan → 4 files
-- Implement → 3-5 phút
-- Assessment → confidence: 0.92, risk: low
-
-**Files created:** `page.tsx`, `Hero.tsx`, `ProductCard.tsx`, `products.ts`
-
----
-
-## 🛒 Day 2: Shopping Cart
-
-```
-/do "Add shopping cart with add/remove functionality"
-```
-
-AI:
-- Agent: fullstack-developer
-- Skills: frontend-development, databases
-- Risk: medium (state management)
-
-**Files created:**
-- `CartContext.tsx`
-- `CartProvider.tsx`
-- `CartButton.tsx`
-- `CartSidebar.tsx`
-
----
-
-## 🔒 Day 3: Authentication
-
-```
-/do "Add user login with email and password"
-```
-
-AI:
-```
-⚠️ Security-Related Feature Detected
-→ Auto-enabling code review
-
-📜 Rules Applied:
-├── Contract P2: Using service layer
-├── Contract S1: No direct DB in routes
-└── Risk Rule: Code review enabled
-```
-
-**Files created:**
-- `AuthService.ts`
-- `auth/route.ts`
-- `LoginForm.tsx`
-- `session.ts`
-
----
-
-## ❓ Day 4: Understand AI Decisions
-
-```
-/why
-```
-
-Output:
-```
-🤖 AGENT SELECTION:
-├── PRIMARY: fullstack-developer
-│   └── WHY: Auth involves frontend + backend
-├── SECONDARY: database-admin
-│   └── WHY: User table schema changes
-
-📚 SKILL ACTIVATION:
-├── ✅ better-auth/ (keyword: "login")
-├── ✅ databases/ (schema modification)
-└── ❌ payment-integration/ (no payment keywords)
-
-📜 RULES APPLIED:
-├── P2: "Use service layer" → AuthService
-└── Risk: "Auth = sensitive" → Enabled review
-```
-
----
-
-## ✅ Day 5: Commit & Deploy
-
-```
-/commit
-```
-
-AI:
-```
-📝 Suggested commit message:
-
-feat: add e-commerce shop with cart and auth
-
-- Add homepage with hero and product grid
-- Add shopping cart with add/remove
-- Add user authentication
-```
-
-```
-/review --security
-```
-
-AI reviews auth implementation → ✅ Passed
-
----
-
-## 📊 Summary
-
-| Metric | Value |
-|--------|-------|
-| Commands used | 7 |
-| Files created | 15+ |
-| Features built | Homepage, Cart, Auth |
-| Code written by hand | ~0 lines |
-| Time spent | ~5 hours total |
-| Reviews triggered | 1 (auto for auth) |
-
----
-
-## 🎓 Key Takeaways
-
-1. **Describe, don't code** - Mô tả muốn gì, AI implement
-2. **AI tự biết giới hạn** - Khi auth, tự enable review
-3. **Explainable** - `/why` giải thích mọi quyết định
-4. **Progressive** - Từ simple → complex tự nhiên
-5. **Safe by default** - Security features auto-protected
-
----
-
-## 💬 User Feedback
-
-> *"Mình chỉ cần mô tả muốn gì, AI tự plan, tự implement, tự biết khi nào cần review. Mình chỉ cần approve và test. Quá tiện!"*
-
----
-
-## 🔗 Related
-
-- [BEGINNER_GUIDE_VI.md](../BEGINNER_GUIDE_VI.md)
-- [QUICK_REFERENCE.md](../QUICK_REFERENCE.md)
-- [docs/ORCHESTRATION.md](./ORCHESTRATION.md)