代码语义搜索与AI助手系统

pip install -r requirements.txt

git clone <your-mcp-repository-url>
cd gitcode_mcp

# Create virtual environment
python -m venv mcp_env

# Activate virtual environment
# Windows:
mcp_env\Scripts\activate
# Linux/Mac:
source mcp_env/bin/activate

pip install -r requirements.txt

# Install Ollama (visit https://ollama.ai)
# After installation, pull the required model:
ollama pull llama3

REPO_NAME = 'your-project-name'
REPO_URL = "git@github.com:username/your-project.git"  # or HTTPS URL
REPO_DIR = "./repo/your-project"
API_KEY = "your-secure-api-key"

# Set API key via environment variable for better security
export MCP_API_KEY="your-secure-api-key"

# Repository Configuration
REPO_NAME = 'your-project-name'           # Name of your project
REPO_URL = "git@github.com:user/repo.git"  # Git repository URL
REPO_DIR = "./repo/your-project"          # Local directory for cloned repo

# Security
API_KEY = "supersecret"                   # API key for server authentication

MODEL_NAME = "all-MiniLM-L6-v2"    # Embedding model
BATCH_EMBED = 256                   # Batch size for embedding
INDEX_WORKERS = max(4, os.cpu_count() // 2)  # Indexing workers
MAX_REQ_PER_MIN = 30               # Rate limiting

# Ensure you're in the project directory and virtual environment is activated
python server.py

[INFO] Repository already exists.
[INFO] 🧠 No embeddings found — running full initial indexing...
[INFO] Embedding 1250 docs...
[INFO] ✅ Indexed 1250 docs.
[INFO] ✅ MCP Server ready on http://0.0.0.0:8000

# Activate virtual environment if needed
python agent.py

🤖 Connected to MCP Code Assistant.

Ask a question (or 'exit'): How is user authentication implemented?

[LLM - Initial MCP Query]
user authentication login session management

================ ROUND 1 =================
[Executing MCP Query] user authentication login session management

✅ Retrieved 8 results from MCP.

🤖 LLM follow-up response: authentication middleware configuration

================ ROUND 2 =================
[Executing MCP Query] authentication middleware configuration

✅ Retrieved 5 results from MCP.

🤖 LLM follow-up response: STOP

🛑 LLM indicated to STOP further searches.

[Final Answer]
Based on the code analysis, user authentication is implemented using...

# Query the codebase
curl -X POST "http://localhost:8000/query" \
  -H "Content-Type: application/json" \
  -H "X-API-Key: supersecret" \
  -d '{"query": "database connection setup", "top_k": 5}'

# Trigger repository sync
curl -X GET "http://localhost:8000/sync" \
  -H "X-API-Key: supersecret"

# Using nohup (Linux/Mac)
nohup python server.py > mcp_server.log 2>&1 &

# Using screen (Linux/Mac)
screen -S mcp-server python server.py

# Using Windows Service or Task Scheduler (Windows)

X-API-Key: your-api-key

{
  "query": "authentication middleware setup",
  "top_k": 10
}

{
  "query": "authentication middleware setup",
  "results": [
    {
      "path": "src/auth/middleware.py",
      "snippet": "class AuthenticationMiddleware:\n    def __init__(self, secret_key):\n        self.secret_key = secret_key\n    \n    def authenticate_request(self, request):\n        token = request.headers.get('Authorization')\n        if not token:\n            raise UnauthorizedError('Missing token')",
      "score": 0.8234
    }
  ]
}

GET /sync?full=false

{
  "message": "Indexed 45 files."
}

import requests

headers = {
    'Content-Type': 'application/json',
    'X-API-Key': 'your-api-key'
}

# Search query
response = requests.post(
    'http://localhost:8000/query',
    json={'query': 'database connection pooling', 'top_k': 5},
    headers=headers
)
results = response.json()

# Manual sync
sync_response = requests.get(
    'http://localhost:8000/sync?full=true',
    headers=headers
)

const headers = {
    'Content-Type': 'application/json',
    'X-API-Key': 'your-api-key'
};

// Search query
const response = await fetch('http://localhost:8000/query', {
    method: 'POST',
    headers: headers,
    body: JSON.stringify({
        query: 'error handling patterns',
        top_k: 8
    })
});
const results = await response.json();

# Search for code
curl -X POST "http://localhost:8000/query" \
  -H "Content-Type: application/json" \
  -H "X-API-Key: your-api-key" \
  -d '{"query": "API rate limiting implementation", "top_k": 3}'

# Force full reindex
curl -X GET "http://localhost:8000/sync?full=true" \
  -H "X-API-Key: your-api-key"

Ask a question (or 'exit'): How does user authentication work in this codebase?

[LLM - Initial MCP Query]
user authentication login verification session management

================ ROUND 1 =================
[Executing MCP Query] user authentication login verification session management

✅ Retrieved 6 results from MCP.

🤖 LLM follow-up response: JWT token validation middleware

================ ROUND 2 =================
[Executing MCP Query] JWT token validation middleware

✅ Retrieved 4 results from MCP.

🤖 LLM follow-up response: STOP

[Final Answer]
Based on the code analysis, user authentication in this codebase works through a multi-layered approach:

1. **JWT Token-Based Authentication**: The system uses JSON Web Tokens stored in the `auth/jwt_handler.py` file...
2. **Middleware Layer**: Authentication middleware in `middleware/auth.py` intercepts requests...
3. **Session Management**: User sessions are managed through...

Ask a question (or 'exit'): Where are database connection settings configured?

[LLM - Initial MCP Query]
database connection configuration settings

================ ROUND 1 =================
[Executing MCP Query] database connection configuration settings

✅ Retrieved 5 results from MCP.

[Final Answer]
Database connection settings are configured in several locations:

1. **Main Configuration**: `config/database.py` contains the primary database connection parameters...
2. **Environment Variables**: The `.env` file includes database credentials...
3. **Connection Pool**: Database pool settings are defined in `utils/db_pool.py`...

import requests

class MCPCodeSearch:
    def __init__(self, api_key, base_url="http://localhost:8000"):
        self.api_key = api_key
        self.base_url = base_url
        self.headers = {
            'Content-Type': 'application/json',
            'X-API-Key': api_key
        }
    
    def search_code(self, query, max_results=5):
        """Search for code snippets related to query"""
        response = requests.post(
            f"{self.base_url}/query",
            json={"query": query, "top_k": max_results},
            headers=self.headers
        )
        return response.json()["results"]
    
    def find_function_usage(self, function_name):
        """Find usages of a specific function"""
        return self.search_code(f"{function_name} function call usage")

# Usage in VS Code extension
mcp = MCPCodeSearch("your-api-key")
results = mcp.find_function_usage("authenticate_user")

def analyze_pull_request(changed_files, mcp_client):
    """Analyze changed files and provide context"""
    analysis = []
    
    for file_path in changed_files:
        # Search for related code patterns
        results = mcp_client.search_code(f"similar patterns {file_path}")
        
        # Find potential dependencies
        deps = mcp_client.search_code(f"imports dependencies {file_path}")
        
        analysis.append({
            "file": file_path,
            "similar_patterns": results[:3],
            "dependencies": deps[:2]
        })
    
    return analysis

def generate_api_docs(mcp_client):
    """Generate API documentation from codebase"""
    
    # Find all API endpoints
    endpoints = mcp_client.search_code("API endpoint route decorator", 20)
    
    # Find authentication methods
    auth = mcp_client.search_code("authentication authorization middleware")
    
    # Find error handling
    errors = mcp_client.search_code("error handling exception status codes")
    
    return {
        "endpoints": endpoints,
        "authentication": auth,
        "error_handling": errors
    }

#!/bin/bash
# .git/hooks/pre-commit

# Trigger MCP reindex before commit
curl -X GET "http://localhost:8000/sync" \
  -H "X-API-Key: ${MCP_API_KEY}" \
  -s > /dev/null

echo "✅ MCP index updated"

# .github/workflows/code-analysis.yml
name: Code Analysis with MCP

on: [push, pull_request]

jobs:
  analyze:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v2
      
      - name: Setup MCP Server
        run: |
          python -m pip install -r requirements.txt
          python server.py &
          sleep 30  # Wait for server to start
      
      - name: Analyze Code Patterns
        run: |
          python scripts/analyze_patterns.py
        env:
          MCP_API_KEY: ${{ secrets.MCP_API_KEY }}

# Secure API key setup
export MCP_API_KEY="your-very-secure-random-key"