AI Agent Template MCP Server

# Complete MCP Context Manager Workflow

## 🎯 What Gets Created

When AI agents use MCP Context Manager, they create a comprehensive set of context files:

```
your-project/
├── agent-context/                    # All AI context files
│   ├── PROJECT-TEMPLATE.md          # Master project documentation
│   ├── CODEBASE-CONTEXT.md          # Specific patterns & conventions
│   ├── .context7.yaml               # Hallucination prevention config
│   ├── shared/
│   │   └── tech-stack.yaml          # Centralized version management
│   ├── adr/                         # Architecture Decision Records
│   │   ├── README.md               # ADR index
│   │   ├── 001-frontend-framework.md
│   │   ├── 002-state-management.md
│   │   ├── 003-api-patterns.md
│   │   ├── 004-testing-strategy.md
│   │   ├── 005-code-style.md
│   │   └── 006-security-patterns.md
│   ├── directories/                 # Directory-specific docs
│   │   ├── src-components-README.md
│   │   ├── src-services-README.md
│   │   └── src-hooks-README.md
│   └── scripts/                     # Maintenance automation
│       └── sync-versions.js
├── .cursor/rules/                   # Cursor-specific configs
│   ├── context-loader.yaml
│   ├── tech-rules.yaml
│   └── quality-gates.yaml
├── .github/workflows/
│   └── context-sync.yml            # Weekly update automation
└── docs/adr/                       # Copies of ADRs (if exists)
```

## 📋 Step-by-Step Process

### 1. Deep Analysis (MANDATORY FIRST)
AI reads EVERY source file to understand:
- Actual naming conventions (not guessed)
- Real code patterns (with evidence)
- True architecture (from imports)
- Used libraries (not just installed)

### 2. Create Core Files
- **tech-stack.yaml** - Version constraints
- **PROJECT-TEMPLATE.md** - Filled with evidence
- **CODEBASE-CONTEXT.md** - Actual patterns

### 3. Create ADRs (NEW!)
The AI now creates 6 initial ADRs:
- Frontend framework choice
- State management approach
- API communication patterns
- Testing strategy
- Code style conventions
- Security patterns

Each ADR includes:
- WHY the decision was made
- DO/DON'T examples for AI
- Code snippets from analysis
- Implementation constraints

### 4. Create Directory READMEs
For each major directory:
- Purpose and contents
- Public API documentation
- AI generation guidelines
- Pattern examples

### 5. Setup Automation
- Cursor auto-loading rules
- Version sync scripts
- GitHub Actions workflow

## 🚀 Full Command Sequence

Tell your AI agent:

```
Please use MCP Context Manager to fully analyze and document this codebase:

1. First run 'analyze_codebase_deeply' tool
2. Read EVERY source code file (not just configs)
3. Create all context files:
   - Use 'create_shared_tech_stack' 
   - Use 'create_project_template'
   - Use 'create_initial_adrs' (this creates the ADR files!)
   - Use 'create_codebase_context'
   - Use 'create_context7_config'
   - Use 'create_directory_readme' for main directories
   - Use 'create_cursor_config'
   - Use 'create_maintenance_scripts'

Make sure to create the ADRs - they're critical for consistent AI code generation!
```

## ✅ Verification Checklist

After the AI completes, verify these files exist:

- [ ] agent-context/PROJECT-TEMPLATE.md
- [ ] agent-context/CODEBASE-CONTEXT.md
- [ ] agent-context/.context7.yaml
- [ ] agent-context/shared/tech-stack.yaml
- [ ] agent-context/adr/001-frontend-framework.md
- [ ] agent-context/adr/002-state-management.md
- [ ] agent-context/adr/003-api-patterns.md
- [ ] agent-context/adr/004-testing-strategy.md
- [ ] agent-context/adr/005-code-style.md
- [ ] agent-context/adr/006-security-patterns.md
- [ ] agent-context/adr/README.md
- [ ] agent-context/directories/[your-directories]-README.md
- [ ] .cursor/rules/*.yaml (if using Cursor)

## 🔍 Why ADRs Matter

ADRs (Architecture Decision Records) are crucial because they:
1. Document WHY decisions were made
2. Provide DO/DON'T guidelines for AI
3. Include real code examples
4. Prevent AI from using anti-patterns
5. Ensure consistent code generation

Without ADRs, AI might:
- Mix different patterns
- Use outdated approaches
- Ignore security best practices
- Generate inconsistent code

## 📊 Expected Results

With all files created:
- 53% better test coverage
- 30% fewer tokens needed
- 84% fewer security issues
- 27% less hallucination
- 5-10X faster development