anchi-toolkit 1.0.0 → 1.2.0

package/.cursor/commands/do.md

@@ -8,6 +8,7 @@ argument-hint: [what you want to do]
 **What:** One command for everything. AI automatically decides the best approach.
 
 ## Usage
+
 ```
 /do "Add user login with Google OAuth"
 /do "Fix the checkout bug"
@@ -18,12 +19,12 @@ argument-hint: [what you want to do]
 
 ## 💰 Model Selection (Cost Optimization)
 
-| Task Type | Recommended Model | Cost | When |
-|-----------|-------------------|------|------|
-| **Simple** (typo, rename) | GPT-5.1 Codex Mini | 💚 Low | Quick fixes |
-| **Medium** (feature, fix) | Sonnet 4.5 | 💛 Medium | Standard dev |
-| **Complex** (architecture) | Opus 4.5 | 🔴 High | Deep reasoning |
-| **Large codebase** (>200 files) | Gemini 3 Flash | 💚 Low | Big context |
+| Task Type                       | Recommended Model  | Cost      | When           |
+| ------------------------------- | ------------------ | --------- | -------------- |
+| **Simple** (typo, rename)       | GPT-5.1 Codex Mini | 💚 Low    | Quick fixes    |
+| **Medium** (feature, fix)       | Sonnet 4.5         | 💛 Medium | Standard dev   |
+| **Complex** (architecture)      | Opus 4.5           | 🔴 High   | Deep reasoning |
+| **Large codebase** (>200 files) | Gemini 3 Flash     | 💚 Low    | Big context    |
 
 ### Auto-Detection Logic
 
@@ -53,13 +54,14 @@ argument-hint: [what you want to do]
 │  └── Is it COMPLEX? → Multi-phase with reviews             │
 │           ↓                                                 │
 │  AUTO WORKFLOW:                                             │
-│  [1] Load context (/memory:load)                           │
+│  [1] Active Memory Load (Context-Aware)                    │
 │  [2] Scout codebase (/scout)                               │
-│  [3] Plan if needed                                        │
-│  [4] Generate boilerplate                                  │
+│  [3] Plan (Parallel Workstreams if needed)                 │
+│  [4] Interactive Check (if --interactive)                  │
 │  [5] Implement                                             │
-│  [6] Fix errors                                            │
+│  [6] Fix errors (Interactive Probing)                      │
 │  [7] Save context (/memory:save)                           │
+│  [8] Post-Mortem (if failure)                              │
 │           ↓                                                 │
 │  ✅ DONE!                                                   │
 │                                                             │
@@ -70,15 +72,17 @@ argument-hint: [what you want to do]
 
 ## Flags (Optional)
 
-| Flag | Effect |
-|------|--------|
-| `--fast` | Skip reviews, quick implementation |
-| `--careful` | Extra validation, more reviews |
-| `--plan-only` | Just create plan, don't implement |
-| `--no-commit` | Don't auto-commit |
-| `--parallel` | Run phases in parallel |
+| Flag            | Effect                                               |
+| --------------- | ---------------------------------------------------- |
+| `--fast`        | Skip reviews, quick implementation                   |
+| `--careful`     | Extra validation, more reviews                       |
+| `--plan-only`   | Just create plan, don't implement                    |
+| `--no-commit`   | Don't auto-commit                                    |
+| `--parallel`    | Run phases in parallel                               |
+| `--interactive` | Pause for user input key steps ([C]ontinue/[M]odify) |
 
 **Examples:**
+
 ```
 /do "quick fix typo" --fast
 /do "payment integration" --careful
@@ -91,58 +95,48 @@ argument-hint: [what you want to do]
 
 AI detects mode from your description:
 
-| Keywords | Mode |
-|----------|------|
-| fix, bug, error, broken, issue | 🔧 Fix Mode |
-| add, create, new, implement, build | ✨ Feature Mode |
+| Keywords                           | Mode             |
+| ---------------------------------- | ---------------- |
+| fix, bug, error, broken, issue     | 🔧 Fix Mode      |
+| add, create, new, implement, build | ✨ Feature Mode  |
 | refactor, optimize, improve, clean | ♻️ Refactor Mode |
-| update, change, modify | 📝 Update Mode |
+| update, change, modify             | 📝 Update Mode   |
 
 ---
 
-## Output
-
-```
-🎯 /do analyzing...
-
-📋 Detected: NEW FEATURE
-📊 Complexity: Medium
-🎯 Approach: Plan → Implement → Test
+## Output Format (Professional Reporting)
 
-═══════════ WORKFLOW ═══════════
+By default, use **Selective Transparency**. Show key roles and decisions, but hide verbose technical steps.
 
-[1/6] 🧠 Loading context...
-      ✅ Found previous session
+### ✅ Success Case
 
-[2/6] 🔍 Scouting codebase...
-      ✅ Found 5 related files
-
-[3/6] 📋 Creating plan...
-      ✅ Plan: 3 files to create, 2 to modify
-
-[4/6] 📝 Generating boilerplate...
-      ✅ Created: LoginPage, AuthAPI, AuthService
-
-[5/6] 💻 Implementing...
-      ✅ Completed: Google OAuth flow
-
-[6/6] 🔧 Fixing issues...
-      ✅ Fixed: 2 TypeScript errors
+```
+🎯 Goal: Implement Google OAuth
+🤖 Orchestration:
+  • Agent: `Fullstack Dev` (+ `Security Auditor` for auth check)
+  • Risk: 🛡️ High (Auth modification) - Approved via policy.
+📝 Execution:
+  • Created `src/services/auth.service.ts`
+  • Updated `src/pages/login.tsx`
+✅ Result: Authentication flow implemented and verified.
+```
 
-═══════════ RESULT ═══════════
+### ⚠️ Risk/Approval Case
 
-✅ /do completed!
+```
+🎯 Task: Delete user database table
+⚠️ RISK: Critical (Data Loss)
+❓ Approval Required: This will wipe production data.
+👉 [Y]es / [N]o ?
+```
 
-📊 Summary:
-  • Created: 3 files
-  • Modified: 2 files
-  • Lines: +250, -30
-  • Time: 8 minutes
+### ❌ Failure Case
 
-📌 Next:
-  • /test - Run tests
-  • /commit - Save changes
-  • /do "next feature" - Continue
+```
+🎯 Task: Fix login bug
+❌ Failed: Could not resolve dependency `better-auth`.
+🔍 Analysis: Package version mismatch.
+💡 Suggestion: Run `npm install better-auth@latest`
 ```
 
 ---
@@ -150,6 +144,7 @@ AI detects mode from your description:
 ## Modes
 
 ### 🔧 Fix Mode
+
 ```
 /do "Fix the login button not working"
 
@@ -160,6 +155,7 @@ AI detects mode from your description:
 ```
 
 ### ✨ Feature Mode
+
 ```
 /do "Add user dashboard"
 
@@ -170,6 +166,7 @@ AI detects mode from your description:
 ```
 
 ### ♻️ Refactor Mode
+
 ```
 /do "Refactor API error handling"
 
@@ -183,11 +180,11 @@ AI detects mode from your description:
 
 ## Preset Integration
 
-| Preset | Default Behavior |
-|--------|------------------|
-| Rapid | --fast (skip reviews) |
-| Professional | Standard + tests |
-| Enterprise | --careful (full reviews) |
+| Preset       | Default Behavior         |
+| ------------ | ------------------------ |
+| Rapid        | --fast (skip reviews)    |
+| Professional | Standard + tests         |
+| Enterprise   | --careful (full reviews) |
 
 ---
 
@@ -195,29 +192,29 @@ AI detects mode from your description:
 
 ### Agents (Auto-Selected Based on Task)
 
-| Task Type | Primary Agent | Support Agents |
-|-----------|---------------|----------------|
-| **Feature** | `fullstack-developer.md` | `planner.md`, `brainstormer.md` |
-| **Fix/Debug** | `debugger.md` | `code-reviewer.md` |
-| **Refactor** | `code-reviewer.md` | `fullstack-developer.md` |
-| **UI Work** | `ui-ux-designer.md` | `design-system-architect.md` |
-| **API Work** | `fullstack-developer.md` | `database-admin.md` |
-| **Complex** | `planner.md` | `researcher.md`, `brainstormer.md` |
+| Task Type     | Primary Agent            | Support Agents                     |
+| ------------- | ------------------------ | ---------------------------------- |
+| **Feature**   | `fullstack-developer.md` | `planner.md`, `brainstormer.md`    |
+| **Fix/Debug** | `debugger.md`            | `code-reviewer.md`                 |
+| **Refactor**  | `code-reviewer.md`       | `fullstack-developer.md`           |
+| **UI Work**   | `ui-ux-designer.md`      | `design-system-architect.md`       |
+| **API Work**  | `fullstack-developer.md` | `database-admin.md`                |
+| **Complex**   | `planner.md`             | `researcher.md`, `brainstormer.md` |
 
 ### Skills (Auto-Activated Based on Tech Detection)
 
-| Detected Tech | Skills Activated |
-|---------------|------------------|
-| React/Next.js | `frontend-development/`, `web-frameworks/` |
-| Node/Express | `backend-development/`, `web-frameworks/` |
-| Prisma/DB | `databases/`, `backend-development/` |
-| Tailwind | `ui-styling/`, `aesthetic/` |
-| TypeScript | `frontend-development/`, `backend-development/` |
-| DDD Pattern | `ddd-modular-monolith/` |
-| Testing | `debugging/`, `code-review/` |
-| AI/ML | `ai-multimodal/`, `claude-code/` |
-| Mobile | `mobile-development/` |
-| 3D/Three.js | `threejs/` |
+| Detected Tech | Skills Activated                                |
+| ------------- | ----------------------------------------------- |
+| React/Next.js | `frontend-development/`, `web-frameworks/`      |
+| Node/Express  | `backend-development/`, `web-frameworks/`       |
+| Prisma/DB     | `databases/`, `backend-development/`            |
+| Tailwind      | `ui-styling/`, `aesthetic/`                     |
+| TypeScript    | `frontend-development/`, `backend-development/` |
+| DDD Pattern   | `ddd-modular-monolith/`                         |
+| Testing       | `debugging/`, `code-review/`                    |
+| AI/ML         | `ai-multimodal/`, `claude-code/`                |
+| Mobile        | `mobile-development/`                           |
+| 3D/Three.js   | `threejs/`                                      |
 
 ### Auto-Activation Logic
 
@@ -230,7 +227,7 @@ AI detects mode from your description:
 │
 [ACTIVATE]
 ├── Agents: fullstack-developer, planner
-├── Skills: 
+├── Skills:
 │   ├── frontend-development/
 │   ├── backend-development/
 │   ├── databases/
@@ -240,11 +237,11 @@ AI detects mode from your description:
 
 ### Mode-Specific Activation
 
-| Flag | Additional Agents | Additional Skills |
-|------|-------------------|-------------------|
-| `--careful` | `code-reviewer.md`, `tester.md` | `code-review/`, `debugging/` |
-| `--fast` | (minimal) | (minimal) |
-| `--plan-only` | `planner.md`, `researcher.md` | `planning/`, `research/` |
+| Flag          | Additional Agents               | Additional Skills            |
+| ------------- | ------------------------------- | ---------------------------- |
+| `--careful`   | `code-reviewer.md`, `tester.md` | `code-review/`, `debugging/` |
+| `--fast`      | (minimal)                       | (minimal)                    |
+| `--plan-only` | `planner.md`, `researcher.md`   | `planning/`, `research/`     |
 
 ---
 
@@ -268,7 +265,7 @@ Total: 40+ commands → 1 command
 
 ```yaml
 do:
-  default_mode: auto      # auto, fast, careful
+  default_mode: auto # auto, fast, careful
   auto_commit: false
   auto_test: true
   show_plan: true
@@ -279,11 +276,43 @@ do:
 
 ---
 
+---
+
+## 🚀 Advanced Capabilities (New)
+
+### 🧠 Active Memory
+
+Instead of loading full context, AI now uses **Context-Aware Retrieval**:
+
+- **/do matches keywords** → Queries `context.json` specifically for related decisions.
+- **Reduces noise** → Faster processing, higher accuracy.
+
+### 🔄 Self-Correction Loop
+
+If a task fails repeatedly (>2 times):
+
+1. **Post-Mortem Chain** activates.
+2. AI analyzes root cause.
+3. Updates `failure-modes.yaml` with a new rule.
+4. Future tasks will AVOID this same mistake.
+
+### 🕵️ Interactive Debugging
+
+Debugger agent now has **Probing** capability:
+
+1. Inject `console.log` probes.
+2. Run test to capture dynamic state.
+3. Analyze output.
+4. Remove probes and apply fix.
+
+---
+
 ## 🔄 Dynamic Orchestration (RUNTIME REQUIRED)
 
 > **AI MUST execute this logic BEFORE implementation**
 
 ### Step 1: Load Config
+
 ```
 READ: .cursor/agent-skill-index.yaml
 READ: .cursor/orchestration.yaml
@@ -291,6 +320,7 @@ READ: .cursor/failure-modes.yaml
 ```
 
 ### Step 2: Analyze Request
+
 ```
 EXTRACT keywords from user description
 IDENTIFY affected file patterns
@@ -298,6 +328,7 @@ DETERMINE complexity (simple/medium/complex)
 ```
 
 ### Step 3: Dynamic Agent Selection
+
 ```
 CHECK command_mapping./do:
   - default: fullstack-developer
@@ -311,6 +342,7 @@ CHECK skill.requires_security_review flags
 ```
 
 ### Step 4: Dynamic Skill Activation
+
 ```
 FOR EACH skill IN skills:
   IF skill.activation.keyword MATCHES request → ACTIVATE
@@ -319,6 +351,7 @@ FOR EACH skill IN skills:
 ```
 
 ### Step 5: Check Failure Modes
+
 ```
 IF confidence < 0.5 → ASK user
 IF security_concern → ESCALATE to security-auditor
@@ -327,6 +360,7 @@ IF outside_expertise → ESCALATE to appropriate agent
 ```
 
 ### Step 6: Determine Risk Level
+
 ```
 SCAN keywords against orchestration.yaml risk_levels:
   - low: ["typo", "rename"] → quick_fix chain
@@ -336,6 +370,7 @@ SCAN keywords against orchestration.yaml risk_levels:
 ```
 
 ### Step 7: Build Execution Chain
+
 ```
 SELECT chain from orchestration.yaml based on risk
 APPLY required approvals
@@ -343,6 +378,7 @@ QUEUE agents in order
 ```
 
 ### Step 8: Log to Audit
+
 ```
 WRITE to .ai-audit/YYYY-MM-DD.jsonl:
 {
@@ -356,6 +392,7 @@ WRITE to .ai-audit/YYYY-MM-DD.jsonl:
 ```
 
 ### Step 9: Execute
+
 ```
 RUN chain with orchestrated agents and skills
 APPLY failure-modes.yaml restraints throughout

package/.cursor/commands/help.md

@@ -29,7 +29,7 @@ argument-hint: [optional: command name]
 ## Output
 
 ```
-📚 anchi-kit v3.0.0 - Command Reference
+📚 anchi-toolkit v1.0.0 - Command Reference
 
 ═══════════ ESSENTIAL (5) ═══════════
 
@@ -55,6 +55,13 @@ argument-hint: [optional: command name]
 /integrate "..."    Third-party integrations
 /memory             Context management
 
+═══════════ META (4) ═══════════
+
+/config             View/edit config
+/audit              AI decision logs
+/migrate            Preset migration
+/why                Explain AI decisions
+
 ═══════════ QUICK START ═══════════
 
 Beginner:
@@ -103,19 +110,19 @@ Examples:
 
 > Help uses minimal AI - static content lookup.
 
-
----
-
-##  Dynamic Orchestration
-
-> **AI MUST read orchestration config before execution**
-
-```
-1. READ .cursor/agent-skill-index.yaml
-2. CHECK command_mapping for this command
-3. APPLY skill activation rules
-4. CHECK .cursor/failure-modes.yaml
-5. LOG decision to .ai-audit/
-```
-
+
+---
+
+##  Dynamic Orchestration
+
+> **AI MUST read orchestration config before execution**
+
+```
+1. READ .cursor/agent-skill-index.yaml
+2. CHECK command_mapping for this command
+3. APPLY skill activation rules
+4. CHECK .cursor/failure-modes.yaml
+5. LOG decision to .ai-audit/
+```
+
  [docs/ORCHESTRATION_RUNTIME.md](../docs/ORCHESTRATION_RUNTIME.md)

package/.cursor/commands/skill-learn.md

@@ -0,0 +1,48 @@
+---
+description: 🧠 Teach AI new skills from the web
+argument-hint: [skill name / tech stack]
+---
+
+# /skill-learn - Infinite Knowledge
+
+**What:** Crawls documentation and creates a new Skill Module dynamically.
+
+## Logic
+
+```
+┌─────────────────────────────────────────────────────────────┐
+│ 🧠 /skill-learn - KNOWLEDGE ACQUISITION                     │
+├─────────────────────────────────────────────────────────────┤
+│                                                             │
+│  [1] ANALYZE REQUEST: "Learn Rust"                          │
+│      ├── Identify Tech: Rust Language                       │
+│      └── Identify Goal: syntax, best practices, packaging   │
+│                                                             │
+│  [2] RESEARCH (Agent: Researcher):                          │
+│      ├── Search Official Docs (rust-lang.org)               │
+│      ├── Search Popular Patterns (Awesome Rust)             │
+│      └── Extract: "Do's & Don'ts", "Snippet Patterns"       │
+│                                                             │
+│  [3] SYNTHESIZE (Agent: Docs-Manager):                      │
+│      ├── Template: .cursor/skills/_template.md             │
+│      ├── Fill: Rules, File Patterns, Dependencies           │
+│      └── Create: .cursor/skills/rust-development/README.md  │
+│                                                             │
+│  [4] REGISTER:                                              │
+│      ├── Append to: .cursor/agent-skill-index.yaml          │
+│      └── Map to: fullstack-developer (or create new role)   │
+│                                                             │
+│  [5] OUTPUT:                                                │
+│      "✅ Learned Skill: Rust Development                    │
+│       Activated for *.rs files."                            │
+│                                                             │
+└─────────────────────────────────────────────────────────────┘
+```
+
+## Usage
+
+```
+/skill-learn "Tailwind CSS"
+/skill-learn "Go Fiber Framework"
+/skill-learn "Solidity Smart Contracts"
+```

package/.cursor/commands/start.md

@@ -1,135 +1,56 @@
 ---
-description: ⚡⚡⚡⚡ Bắt đầu sử dụng anchi-kit - Entry point cho mọi dự án
+description: 🚀 Start/Resume project with Context Recall
+argument-hint: [optional: new project name]
 ---
 
-# INSTRUCTION: Follow this workflow exactly
+# /start - Project Intelligence
 
-> ⛔ **CRITICAL: PAUSE POINTS**
-> 
-> This command has MULTIPLE decision points where you MUST STOP.
-> 
-> At each decision point:
-> 1. **ONLY show the question** - No other actions
-> 2. **DO NOT call any tools** in parallel with the question
-> 3. **WAIT for user's number input** before proceeding
-> 4. **DO NOT assume or decide** for the user
->
-> Decision points:
-> - STEP 1: Ask "1 or 2" (existing vs new project) → STOP
-> - STEP 2B: Ask "1, 2, or 3" (preset selection) → STOP
+**What:** Initialize project or RESUME work with context awareness.
 
----
-
-## ▶️ STEP 1: Show Welcome (⛔ STOP HERE)
-
-**Show ONLY this, then STOP:**
-
-```
-👋 Chào mừng đến với anchi-kit!
-
-Tôi sẽ giúp bạn setup dự án. Chỉ cần trả lời 1 câu hỏi:
-
-📂 Dự án của bạn là:
-  1️⃣ Dự án có sẵn - Đã có code
-  2️⃣ Dự án mới - Bắt đầu từ đầu
-
-👉 Gõ 1 hoặc 2:
-```
-
-⛔ **STOP. Wait for user to type 1 or 2. Do not proceed until user responds.**
-
----
-
-## ▶️ STEP 2A: Existing Project (User typed "1")
-
-If user typed 1, proceed with scanning:
-
-### Phase 1: Deploy Scout Agents
-Spawn scout agents to scan codebase in parallel.
-
-### Phase 2: Activate Skills
-Based on detected tech stack.
-
-### Phase 3: Auto-Detect Preset
-Analyze and suggest preset.
-
-### Phase 4: Fill Docs
-Generate project documentation.
-
----
-
-## ▶️ STEP 2B: New Project (User typed "2") (⛔ STOP HERE)
-
-If user typed 2, **show ONLY this, then STOP:**
+## Logic (The "Brain")
 
 ```
-📂 Bạn chọn: 2️⃣ Dự án mới
-
-📋 Chọn Architecture Preset:
-
 ┌─────────────────────────────────────────────────────────────┐
-│ 1️⃣ 🚀 RAPID MVP                                             │
-│    SQLite | Flat structure | Skip tests OK                 │
-│    Best for: Prototype, demo, hackathon                    │
-│    ⏱️ ~5 min setup                                          │
+│ 🧠 /start - PROJECT RECALL                                  │
 ├─────────────────────────────────────────────────────────────┤
-│ 2️⃣ 💼 PROFESSIONAL                                          │
-│    PostgreSQL | Modular | Tests required                   │
-│    Best for: Startup, production app                       │
-│    ⏱️ ~10 min setup                                         │
-├─────────────────────────────────────────────────────────────┤
-│ 3️⃣ 🏢 ENTERPRISE                                            │
-│    PostgreSQL + Redis | DDD | Full compliance              │
-│    Best for: Large project, regulated industry             │
-│    ⏱️ ~20 min setup                                         │
+│                                                             │
+│  [1] CHECK: Is this a new project?                          │
+│      ├── YES → Show Preset Selection (Rapid/Pro/Enterprise) │
+│      └── NO  → Activate RECALL MODE                         │
+│                                                             │
+│  [2] RECALL MODE (The "Wow" Factor):                        │
+│      ├── READ: .ai-memory/feature_status.json              │
+│      ├── READ: .ai-audit/latest.jsonl                       │
+│      └── ANALYZE: What was the last active task?            │
+│                                                             │
+│  [3] OUTPUT:                                                │
+│      "Welcome back! 👋                                      │
+│       Last time we were working on [Task Name].             │
+│       Status: [In Progress/Done]                            │
+│       Files: [Active Files]                                 │
+│                                                             │
+│       👉 Continue? [Y]es / [C]hange Task"                   │
+│                                                             │
 └─────────────────────────────────────────────────────────────┘
-
-👉 Gõ 1, 2, hoặc 3:
 ```
 
-⛔ **STOP. Wait for user to type 1, 2, or 3. Do not proceed until user responds.**
+## Setup Steps (New Project)
 
----
+If folder is empty:
 
-## ▶️ STEP 3: Execute Preset Setup
+1. Detect Tech Stack (Node, Python, Go...)
+2. Suggest Best Preset
+3. Run `npx anchi-toolkit init`
 
-After user selects preset (1/2/3):
+## Memory Schema (For Recall)
 
-1. Create project folder if needed
-2. Initialize project with selected preset
-3. Copy preset-specific .cursorrules
-4. Setup documentation templates
-5. Show completion message
-
----
-
-## Agent & Skill Summary
-
-| Agent | Role |
-|-------|------|
-| `scout` | Scan codebase |
-| `researcher` | Deep analysis |
-| `planner` | Preset detection |
-| `docs-manager` | Fill templates |
-
-| Detected Tech | Skills |
-|---------------|--------|
-| Next.js/React | `frontend-development/` |
-| Prisma/DB | `databases/` |
-| TailwindCSS | `ui-styling/` |
-
----
-
-## Dynamic Orchestration
-
-> **AI MUST read orchestration config before execution**
+AI looks for `.ai-memory/session.json`:
 
+```json
+{
+  "last_active": "2024-12-27T10:00:00Z",
+  "current_task": "Dashboard UI Implementation",
+  "active_files": ["src/dashboard.tsx", "src/api.ts"],
+  "next_step": "Implement chart component"
+}
 ```
-1. READ .cursor/agent-skill-index.yaml
-2. CHECK command_mapping for this command
-3. APPLY skill activation rules
-4. CHECK .cursor/failure-modes.yaml
-5. LOG decision to .ai-audit/
-```
-
-📖 [docs/ORCHESTRATION_RUNTIME.md](../docs/ORCHESTRATION_RUNTIME.md)