Other MCP Clients

PromptRepo implements the Model Context Protocol (MCP) specification, making it compatible with any MCP client—not just Claude Desktop and Claude Code.

What is MCP?

The Model Context Protocol is an open standard for connecting AI applications to external data sources and tools. It uses JSON-RPC 2.0 over HTTP/HTTPS, allowing any MCP-compatible client to:

Discover available tools and resources
Invoke tools with structured parameters
Receive typed responses

PromptRepo MCP Endpoint

All MCP clients connect to the same endpoint:

POST https://your-app-url/api/mcp
Content-Type: application/json

Authentication

PromptRepo supports two authentication headers:

Authorization: Bearer YOUR_API_KEY

Anonymous Access

If no API key is provided, the endpoint returns public prompts only. This is useful for:

Sharing prompt libraries publicly
Demo environments
Read-only integrations

Available Tools

PromptRepo exposes four MCP tools:

1. `list_prompts`

Returns all prompts accessible to the authenticated user. Request:

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "list_prompts",
  "params": {}
}

Response:

{
  "jsonrpc": "2.0",
  "id": 1,
  "result": {
    "prompts": [
      {
        "id": "abc123",
        "name": "code-review",
        "content": "Review this {{language}} code for...",
        "variables": ["language"],
        "tags": ["development", "quality"],
        "is_public": false
      }
    ]
  }
}

2. `get_prompt`

Fetches a specific prompt by ID. Request:

{
  "jsonrpc": "2.0",
  "id": 2,
  "method": "get_prompt",
  "params": {
    "prompt_id": "abc123"
  }
}

Response:

{
  "jsonrpc": "2.0",
  "id": 2,
  "result": {
    "id": "abc123",
    "name": "code-review",
    "content": "Review this {{language}} code for...",
    "variables": ["language"],
    "tags": ["development", "quality"],
    "is_public": false,
    "created_at": "2026-02-15T10:30:00Z",
    "updated_at": "2026-02-20T14:45:00Z"
  }
}

3. `resolve_prompt`

Fetches a prompt and substitutes {{variable}} placeholders with provided values. Request:

{
  "jsonrpc": "2.0",
  "id": 3,
  "method": "resolve_prompt",
  "params": {
    "prompt_id": "abc123",
    "variables": {
      "language": "TypeScript"
    }
  }
}

Response:

{
  "jsonrpc": "2.0",
  "id": 3,
  "result": {
    "original": "Review this {{language}} code for...",
    "resolved": "Review this TypeScript code for...",
    "variables_used": {
      "language": "TypeScript"
    }
  }
}

4. `search_prompts`

Full-text search across prompt names, content, and tags. Request:

{
  "jsonrpc": "2.0",
  "id": 4,
  "method": "search_prompts",
  "params": {
    "query": "code review typescript"
  }
}

Response:

{
  "jsonrpc": "2.0",
  "id": 4,
  "result": {
    "prompts": [
      {
        "id": "abc123",
        "name": "code-review",
        "content": "Review this {{language}} code...",
        "relevance_score": 0.92
      }
    ],
    "total": 1
  }
}

Error Handling

All errors use JSON-RPC 2.0 error codes:

Code	Meaning	Common Cause
`-32700`	Parse error	Invalid JSON in request body
`-32600`	Invalid Request	Missing required fields
`-32601`	Method not found	Unsupported tool name
`-32602`	Invalid params	Wrong parameter types
`4001`	Invalid API key	Key is incorrect or revoked
`4004`	Not found	Prompt ID doesn’t exist
`5000`	Internal error	Server-side issue

Example error response:

{
  "jsonrpc": "2.0",
  "id": 5,
  "error": {
    "code": 4001,
    "message": "Invalid API key."
  }
}

Compatible Clients

Cursor

Cursor (the AI-powered code editor) supports MCP servers. Configuration (~/.cursor/mcp_config.json):

{
  "mcpServers": {
    "promptrepo": {
      "url": "https://your-app-url/api/mcp",
      "headers": {
        "x-api-key": "YOUR_API_KEY"
      }
    }
  }
}

Restart Cursor after saving the configuration.

Continue.dev

Continue (VS Code extension) supports MCP. Configuration (~/.continue/config.json):

{
  "mcp": [
    {
      "name": "promptrepo",
      "url": "https://your-app-url/api/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  ]
}

Custom MCP Clients

You can build custom integrations using any HTTP client. Example (Node.js with fetch):

const mcpClient = {
  url: 'https://your-app-url/api/mcp',
  apiKey: 'YOUR_API_KEY',
  
  async call(method, params = {}) {
    const response = await fetch(this.url, {
      method: 'POST',
      headers: {
        'Content-Type': 'application/json',
        'x-api-key': this.apiKey,
      },
      body: JSON.stringify({
        jsonrpc: '2.0',
        id: Date.now(),
        method,
        params,
      }),
    });
    
    const result = await response.json();
    
    if (result.error) {
      throw new Error(result.error.message);
    }
    
    return result.result;
  },
};

// Usage
const prompts = await mcpClient.call('list_prompts');
const resolved = await mcpClient.call('resolve_prompt', {
  prompt_id: 'abc123',
  variables: { language: 'Python' },
});

Example (Python):

import requests
import json

class MCPClient:
    def __init__(self, url, api_key):
        self.url = url
        self.api_key = api_key
        self.request_id = 0
    
    def call(self, method, params=None):
        self.request_id += 1
        
        response = requests.post(
            self.url,
            headers={
                'Content-Type': 'application/json',
                'x-api-key': self.api_key,
            },
            json={
                'jsonrpc': '2.0',
                'id': self.request_id,
                'method': method,
                'params': params or {},
            },
        )
        
        result = response.json()
        
        if 'error' in result:
            raise Exception(result['error']['message'])
        
        return result['result']

# Usage
client = MCPClient(
    url='https://your-app-url/api/mcp',
    api_key='YOUR_API_KEY',
)

prompts = client.call('list_prompts')
resolved = client.call('resolve_prompt', {
    'prompt_id': 'abc123',
    'variables': {'language': 'Python'},
})

CORS Support

The PromptRepo MCP endpoint includes CORS headers:

Access-Control-Allow-Origin: *
Access-Control-Allow-Methods: POST, OPTIONS
Access-Control-Allow-Headers: Content-Type, Authorization, x-api-key

This allows browser-based MCP clients to connect without CORS errors. Preflight request example:

curl -X OPTIONS https://your-app-url/api/mcp \
  -H "Access-Control-Request-Method: POST" \
  -H "Access-Control-Request-Headers: x-api-key"

Testing the Endpoint

Use curl to test the MCP endpoint:

List Prompts

curl -X POST https://your-app-url/api/mcp \
  -H "Content-Type: application/json" \
  -H "x-api-key: YOUR_API_KEY" \
  -d '{
    "jsonrpc": "2.0",
    "id": 1,
    "method": "list_prompts",
    "params": {}
  }'

Search Prompts

curl -X POST https://your-app-url/api/mcp \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{
    "jsonrpc": "2.0",
    "id": 2,
    "method": "search_prompts",
    "params": {
      "query": "typescript"
    }
  }'

Resolve Prompt

curl -X POST https://your-app-url/api/mcp \
  -H "Content-Type: application/json" \
  -H "x-api-key: YOUR_API_KEY" \
  -d '{
    "jsonrpc": "2.0",
    "id": 3,
    "method": "resolve_prompt",
    "params": {
      "prompt_id": "abc123",
      "variables": {
        "language": "Python"
      }
    }
  }'

Self-Hosting Considerations

If you’re self-hosting PromptRepo:

Environment Variables

Ensure SUPABASE_SERVICE_ROLE_KEY is set in your deployment environment:

# Vercel
vercel env add SUPABASE_SERVICE_ROLE_KEY

# Railway
railway variables set SUPABASE_SERVICE_ROLE_KEY=your-key

# Docker
docker run -e SUPABASE_SERVICE_ROLE_KEY=your-key ...

Middleware Exclusion

The MCP endpoint (/api/mcp) is automatically excluded from session authentication middleware, allowing API key-based auth to work without browser cookies. If you modify src/middleware.ts, ensure /api/mcp remains in the exclusion list:

const publicPaths = [
  '/auth',
  '/p/',      // Public sharing links
  '/api/mcp', // MCP endpoint (uses API key auth)
];

Rate Limiting

For production deployments, consider adding rate limiting to the MCP endpoint:

// src/app/api/mcp/route.ts
import { ratelimit } from '@/lib/ratelimit';

export async function POST(request: Request) {
  const identifier = request.headers.get('x-api-key') || 'anonymous';
  const { success } = await ratelimit.limit(identifier);
  
  if (!success) {
    return jsonResponse(
      { jsonrpc: '2.0', id: null, error: { code: 429, message: 'Too many requests' } },
      429
    );
  }
  
  // ... rest of handler
}

Security Best Practices

API Key Security:

Store keys in environment variables or secret managers
Never commit keys to version control
Rotate keys regularly
Use separate keys per environment/client
Revoke compromised keys immediately from /profile

Endpoint Security:

Always use HTTPS in production
Monitor API usage for anomalies
Implement rate limiting to prevent abuse
Validate all input parameters
Log authentication failures for auditing

Setup

Available Tools

AI Agent Integration

What is MCP?

PromptRepo MCP Endpoint

Authentication

Anonymous Access

Available Tools

1. `list_prompts`

2. `get_prompt`

3. `resolve_prompt`

4. `search_prompts`

Error Handling

Compatible Clients

Cursor

Continue.dev

Custom MCP Clients

CORS Support

Testing the Endpoint

List Prompts

Search Prompts

Resolve Prompt

Self-Hosting Considerations

Environment Variables

Middleware Exclusion

Rate Limiting

Security Best Practices

Next Steps

Build docs developers (and LLMs) love

Setup

Available Tools

AI Agent Integration

​What is MCP?

​PromptRepo MCP Endpoint

​Authentication

​Anonymous Access

​Available Tools

​1. list_prompts

​2. get_prompt

​3. resolve_prompt

​4. search_prompts

​Error Handling

​Compatible Clients

​Cursor

​Continue.dev

​Custom MCP Clients

​CORS Support

​Testing the Endpoint

​List Prompts

​Search Prompts

​Resolve Prompt

​Self-Hosting Considerations

​Environment Variables

​Middleware Exclusion

​Rate Limiting

​Security Best Practices

​Next Steps

Build docs developers (and LLMs) love

What is MCP?

PromptRepo MCP Endpoint

Authentication

Anonymous Access

Available Tools

1. `list_prompts`

2. `get_prompt`

3. `resolve_prompt`

4. `search_prompts`

Error Handling

Compatible Clients

Cursor

Continue.dev

Custom MCP Clients

CORS Support

Testing the Endpoint

List Prompts

Search Prompts

Resolve Prompt

Self-Hosting Considerations

Environment Variables

Middleware Exclusion

Rate Limiting

Security Best Practices

Next Steps