API.md•2.04 kB
# API Reference
Programmatic usage of MCPhy in your Node.js projects.
## Installation
```bash
npm install mcphy
```
## Basic Usage
```typescript
import { MCPServer, SwaggerAPIParser, ManifestGenerator } from 'mcphy';
// Parse API specification
const apiSpec = await SwaggerAPIParser.parse('./swagger.yaml');
// Generate MCP manifest
const manifest = await ManifestGenerator.generateFromSwagger(apiSpec);
// Create and start server
const server = new MCPServer({
port: 3000,
manifest,
openaiApiKey: process.env.OPENAI_API_KEY,
apiBaseUrl: 'http://localhost:8000'
});
await server.start();
console.log('🚀 Server running on port 3000');
```
## Query API
```typescript
import { QueryMatcher } from 'mcphy';
const matcher = new QueryMatcher(manifest, openaiApiKey);
const result = await matcher.matchQuery('get all users');
console.log(result);
// {
// endpoint: '/users',
// method: 'GET',
// params: {},
// confidence: 0.9,
// reasoning: 'Matched based on keywords'
// }
```
## REST Endpoints
| Endpoint | Method | Description |
|----------|--------|-------------|
| `/mcp/query` | POST | Natural language query matching |
| `/api/proxy/*` | ALL | Proxy requests to your API |
| `/api/endpoints` | GET | List all available endpoints |
| `/health` | GET | Health check |
## Examples
### Express Integration
```typescript
import express from 'express';
import { MCPServer } from 'mcphy';
const app = express();
const mcpServer = new MCPServer({
port: 3001,
manifest: myManifest,
apiBaseUrl: 'https://api.example.com'
});
// Add custom middleware
app.use('/api', mcpServer.getApp());
app.listen(3000);
```
### Custom Query Processing
```typescript
import { QueryMatcher } from 'mcphy';
const matcher = new QueryMatcher(manifest);
// Process multiple queries
const queries = [
'get all users',
'create new product',
'delete order 123'
];
for (const query of queries) {
const result = await matcher.matchQuery(query);
console.log(`${query} → ${result.method} ${result.endpoint}`);
}
```