ClosedLoop MCP Server

A lightweight Model Context Protocol (MCP) server that provides AI clients with access to ClosedLoop customer feedback data and advanced search capabilities.

Features

• MCP Protocol Support: Full MCP protocol implementation for AI client integration
• API Key Authentication: Secure team-based API key authentication
• Team Isolation: All data access is scoped to the team that owns the API key
• Advanced Search: Full-text search with relevance ranking across 8+ insight fields
• Three Core Tools:
  • list_insights: Get customer insights with date range and pagination
  • get_insight_detail: Get detailed information about specific insights
  • search_insights: NEW Advanced full-text search with filtering and relevance ranking

Installation

npm install -g @closedloop-ai/mcp-client

Configuration

Set the following environment variables:

• CLOSEDLOOP_API_KEY: Your ClosedLoop API key (required)
• CLOSEDLOOP_SERVER_URL: Server URL (default: https://mcp.closedloop.sh)

Usage with Claude

Add this configuration to your Claude Desktop app:

{
  "mcpServers": {
    "closedloop": {
      "command": "npx",
      "args": ["@closedloop-ai/mcp-client"],
      "env": {
        "CLOSEDLOOP_API_KEY": "your-api-key-here",
        "CLOSEDLOOP_SERVER_URL": "https://mcp.closedloop.sh"
      }
    }
  }
}

MCP Tools

list_insights

Retrieve customer insights with optional filtering and pagination.

Parameters:

• date_from (optional): Start date (YYYY-MM-DD)
• date_to (optional): End date (YYYY-MM-DD)
• page (optional): Page number (default: 1)
• limit (optional): Items per page (default: 20, max: 100)

get_insight_detail

Get detailed information about a specific insight item.

Parameters:

• insight_id (required): UUID of the insight item

search_insights ⭐ NEW

Advanced full-text search across customer insights with relevance ranking and comprehensive filtering.

Key Features:

• Full-text search across 8+ insight fields
• Relevance ranking - most relevant results appear first
• Phrase search and fuzzy matching
• Field-specific search - search specific fields or all fields
• Advanced filtering - category, severity, date range, source
• Language agnostic - works with any language

Parameters:

• query (required): Search query text
• fields (optional): Specific fields to search in
• category (optional): Filter by insight category
• severity (optional): Filter by severity level
• date_from (optional): Start date filter
• date_to (optional): End date filter
• source (optional): Filter by data source
• page (optional): Page number (default: 1)
• limit (optional): Items per page (default: 20, max: 100)

Searchable Fields:

• signal_title - Insight title
• content - Raw feedback content
• pain_point - Customer pain points
• workaround - Customer workarounds
• use_case - Use case descriptions
• feature_area - Feature categorization
• competitor_gap - Competitive analysis
• willingness_to_pay - Payment willingness info

Example Usage

Once configured, you can ask Claude questions like:

Basic Insights

• "Show me all negative insights from last week"
• "What are the top customer insights this month?"

Advanced Search

• "Search for mobile app performance issues"
• "Find insights about slow loading times in the Performance Issue category"
• "Search for pricing concerns in pain points and willingness to pay fields"
• "Find all high severity bugs from this month"
• "Search for integration problems with Typeform"

Search Examples

• Phrase search: "Find insights about 'mobile app startup'"
• Field-specific: "Search pain points for 'frustrating experience'"
• Category filter: "Find all Feature Request insights about 'real-time collaboration'"
• Severity filter: "Search for critical issues with 'security'"
• Date range: "Find performance issues from January 2024"

License

MIT