MCP Server Documentation

Getting Started

The ClosedLoop AI MCP Server provides AI clients with access to feedback data and insights through the Model Context Protocol. This server enables AI assistants to interact with your ClosedLoop AI data seamlessly.

What is MCP?

The Model Context Protocol (MCP) is a standard for AI clients to connect to data sources and tools. It allows AI assistants to access your ClosedLoop AI feedback data, analyze insights, and provide intelligent responses.

Base URL

Production: https://mcp.closedloop.sh

Quick Start

1. Get your API key

2. Connect your MCP client

mcp://mcp.closedloop.sh

Authorization: Bearer YOUR_API_KEY

3. Start using tools

Your AI client can now access feedback data and provide intelligent insights.

Authentication

All MCP requests require authentication using a Bearer token. Include your API key in the Authorization header.

Authorization: Bearer YOUR_API_KEY

API Key Security

Keep your API key secure and never share it publicly. Rotate keys regularly for enhanced security.

MCP Tools

The ClosedLoop AI MCP Server provides several tools for AI clients to interact with your feedback data and chat functionality.

Feedback Tools

Access and analyze customer feedback data with AI-powered insights.

Chat Tools

Manage conversations and process messages with AI analysis.

Feedback Tools

Tools for accessing and analyzing customer feedback data.

list_feedbacks

Tool: list_feedbacks

Description: Get customer feedback with date range and pagination

Parameters

{
  "date_from": "2024-01-01",    // Optional: Start date (YYYY-MM-DD)
  "date_to": "2024-12-31",      // Optional: End date (YYYY-MM-DD)
  "page": 1,                    // Optional: Page number (default: 1)
  "limit": 20                   // Optional: Items per page (default: 20, max: 100)
}

get_feedback_detail

Tool: get_feedback_detail

Description: Get detailed information about a specific feedback item

Parameters

{
  "feedback_id": "uuid-string"  // Required: UUID of the feedback item
}

Chat Tools

Tools for managing conversations and processing messages with AI analysis.

list_conversations

Tool: list_conversations

Description: Get user conversations for chat interface

Parameters

{
  "limit": 3  // Optional: Number of conversations (default: 3, max: 50)
}

create_conversation

Tool: create_conversation

Description: Create a new chat conversation

Parameters

{
  "title": "Conversation Title"  // Required: Conversation title
}

send_message

Tool: send_message

Description: Send a message in a conversation

Parameters

{
  "conversation_id": "uuid-string",  // Required: UUID of the conversation
  "content": "Message content"        // Required: Message content
}

get_conversation_messages

Tool: get_conversation_messages

Description: Get messages from a conversation

Parameters

{
  "conversation_id": "uuid-string",  // Required: UUID of the conversation
  "limit": 50                        // Optional: Number of messages (default: 50, max: 100)
}

Integration

The ClosedLoop AI MCP Server supports both HTTP and Stdio transport methods for MCP clients.

HTTP Transport

POST https://mcp.closedloop.sh

Content-Type: application/json

Authorization: Bearer YOUR_API_KEY

Stdio Transport

stdin/stdout communication

Authorization header in request meta

MCP Protocol Methods

tools/list - Get available tools

tools/call - Execute a tool with parameters

Examples

List Recent Feedback

Method: tools/call

Tool: list_feedbacks

Parameters:

{

"date_from": "2024-01-01",

"limit": 10

}

Create Chat Conversation

Method: tools/call

Tool: create_conversation

Parameters:

{

"title": "Product Feedback Analysis"

}

Send Message

Method: tools/call

Tool: send_message

Parameters:

{

"conversation_id": "uuid-string",

"content": "Analyze the latest customer feedback"

}