Looply AI - MCP Server

Expose your Looply AI call center data to external AI agents (Claude Desktop, Manus, Cursor, etc.) via the Model Context Protocol.

Installation

# Run directly with npx (no install needed)
npx looply-mcp-server

# Or install globally
npm install -g looply-mcp-server
looply-mcp-server

Quick Start

1. Get your API Key

Go to Settings → API Keys in your Looply dashboard at looplyauto.com and create a new key.

2. Set Environment Variables

export DATABASE_URL="your-database-url"
export LOOPLY_API_KEY="your-api-key"

3. Configure Claude Desktop

Add to ~/Library/Application Support/Claude/claude_desktop_config.json:

{
  "mcpServers": {
    "looply": {
      "command": "npx",
      "args": ["looply-mcp-server"],
      "env": {
        "DATABASE_URL": "your-database-url",
        "LOOPLY_API_KEY": "your-api-key"
      }
    }
  }
}

4. Configure Manus

Add Looply as an MCP server in Manus settings with the same environment variables.

5. Configure Cursor

Add to .cursor/mcp.json in your project:

{
  "mcpServers": {
    "looply": {
      "command": "npx",
      "args": ["looply-mcp-server"],
      "env": {
        "DATABASE_URL": "your-database-url",
        "LOOPLY_API_KEY": "your-api-key"
      }
    }
  }
}

6. Start Using

Ask your AI agent: "Show me today's call summary from Looply"

Available Tools (22)

Calls (3 tools)

| Tool | Description | |------|-------------| | list_calls | List recent calls with filters (status, direction, qualification) | | get_call | Get call details with transcript and AI summary | | search_calls | Search calls by name, phone, or intent keyword |

Contacts (3 tools)

| Tool | Description | |------|-------------| | list_contacts | List CRM contacts with search | | create_contact | Create a new contact | | get_contact | Get contact details |

Leads (3 tools)

| Tool | Description | |------|-------------| | list_leads | List pipeline leads with stage/priority filters | | create_lead | Create a new lead | | update_lead | Update lead stage, priority, or value |

Analytics (2 tools)

| Tool | Description | |------|-------------| | get_dashboard_stats | Dashboard KPIs: calls, answer rate, leads, pipeline value | | get_call_analytics | Detailed analytics: trends, sentiment, qualification breakdown |

Scripts (3 tools)

| Tool | Description | |------|-------------| | list_scripts | List AI call scripts | | get_script | Get script with all steps | | update_script | Update script name, description, or active status |

Callbacks (3 tools)

| Tool | Description | |------|-------------| | list_callbacks | List pending callbacks sorted by priority | | create_callback | Schedule a new callback | | complete_callback | Mark callback as completed |

Configuration (5 tools)

| Tool | Description | |------|-------------| | get_business_hours | Get business hours configuration | | update_business_hours | Update business hours and after-hours action | | list_routing_rules | List smart routing rules | | list_qualification_rules | List AI qualification rules | | get_organization | Get organization profile | | get_usage_summary | Get current billing period usage |

Resources

| Resource | URI | Description | |----------|-----|-------------| | Organization Overview | looply://organization/overview | High-level org stats for AI context |

Authentication

The MCP server authenticates using API keys created in the Looply dashboard. Keys are SHA-256 hashed and matched against the api_keys table. Each key is scoped to a specific organization.

SOVR Audit Integration

All MCP tool calls are audited through the SOVR gate_check layer. Every operation is logged with:

• trace_id: Unique identifier for the operation chain
• tool_name: Which MCP tool was invoked
• parameters: Input parameters (sensitive data redacted)
• decision: allow / block / escalate
• timestamp: When the operation occurred

View audit logs at Settings → Audit Log in your Looply dashboard.

Example Prompts

Once connected, try these with Claude or any MCP-compatible AI:

• "What were the top 5 calls today? Show me the hot leads."
• "Create a callback for contact #12 tomorrow at 2pm, high priority."
• "Show me the sentiment breakdown for this week's calls."
• "List all leads in the proposal stage worth over $10,000."
• "What's our answer rate this month?"
• "Update the AI greeting script to mention our holiday hours."

Development

# Clone the repo
git clone https://github.com/xie38388/looply-clone.git
cd looply-clone/mcp-server

# Install dependencies
npm install

# Run in development mode
DATABASE_URL="..." LOOPLY_API_KEY="..." npm run dev

# Build for distribution
npm run build

# Test with MCP Inspector
npx @modelcontextprotocol/inspector npx tsx src/index.ts

Publishing

# Login to npm
npm login

# Publish
npm publish

License

MIT