Prerak CRM MCP Server

About Prerak CRM

Prerak is a modern, AI-native CRM platform built for sales teams who want to work smarter. It seamlessly integrates contacts, companies, deals, tasks, meetings, and email into a unified workspace, with built-in AI agents that automate pipeline management, forecasting, and customer intelligence. Designed for productivity, Prerak helps teams close deals faster while maintaining deep customer relationships.

Available Tools

The Prerak CRM MCP server exposes 30+ tools across 8 domains:

Contacts (4 tools)

• contacts_list - Retrieve a list of contacts with filtering and pagination support
• contacts_get - Get detailed information about a specific contact
• contacts_create - Create a new contact with email, phone, and custom properties
• contacts_update_properties - Update contact properties like name, email, phone, and custom fields

Companies (5 tools)

• companies_list - Retrieve a list of companies with filtering and pagination
• companies_get - Get detailed information about a specific company
• companies_create - Create a new company with name and custom properties
• companies_list_people - List all contacts associated with a specific company
• companies_update_properties - Update company properties like industry, revenue, and custom fields

Deals (5 tools)

• deals_list - Retrieve a list of deals with status and stage filtering
• deals_get - Get detailed information about a specific deal
• deals_create - Create a new deal with name, amount, stage, and owner assignment
• deals_update - Update deal properties including stage, amount, and custom fields
• deals_delete - Delete a deal from the pipeline

Tasks (6 tools)

• tasks_list - Retrieve a list of tasks with status and priority filtering
• tasks_get - Get detailed information about a specific task
• tasks_create - Create a new task with title, due date, and assignee
• tasks_update - Update task properties including status, priority, and due date
• tasks_delete - Delete a task
• tasks_add_comment - Add a comment to a task for collaboration

Activity (2 tools)

• activity_list - Retrieve activity log entries for CRM records with filtering by date range and type
• activity_create_note - Create activity notes attached to contacts, companies, or deals

AI Agents (4 tools)

• agent_list - List available AI agents configured in your workspace
• agent_run - Execute an AI agent with specified parameters and inputs
• agent_status - Check the status and results of a running agent
• agent_configure - Configure or update agent settings and behaviors

Meetings (4 tools)

• meetings_list - Retrieve a list of meetings with date and attendee filtering
• meetings_get - Get detailed information about a specific meeting
• meetings_summary - Generate an AI summary of meeting notes
• meetings_transcript - Retrieve the transcript of a meeting

Email (4 tools)

• emails_list - Retrieve a list of emails with filtering by sender, recipient, and date
• emails_get - Get detailed information about a specific email
• emails_send - Send an email (with confirmation pattern to prevent accidental sends)
• emails_reply - Reply to an email thread (with confirmation pattern)

Setup Instructions

Prerequisites

1. Prerak backend running (default: http://localhost:3000)
2. Workspace API key from Settings > Workspace > Apps (/settings/apps)

Environment Variables

Configure the following environment variables in your MCP client:

• CRM_MCP_BASE_URL - Backend API URL (default: http://localhost:3000)
• CRM_MCP_AUTH_MODE - Authentication mode: api_key, bearer, or basic (default: api_key)
• CRM_MCP_API_KEY - API key (required when using api_key mode)
• CRM_MCP_TIMEOUT_MS - Request timeout in milliseconds (default: 15000)

Authentication Modes

• api_key - Workspace API key (recommended). Set CRM_MCP_API_KEY to your workspace key.
• bearer - Bearer token authentication. Set CRM_MCP_API_KEY to your bearer token.
• basic - HTTP Basic authentication. Set CRM_MCP_API_KEY to base64-encoded username:password.

Running Locally

Start the server:

pnpm --filter @prerak/mcp-server start

Or with watch mode for development:

pnpm --filter @prerak/mcp-server dev

Validate configuration without opening the stdio transport:

CRM_MCP_VALIDATE_ONLY=1 CRM_MCP_API_KEY=prk_test pnpm --filter @prerak/mcp-server start

Claude Desktop Integration

Add the following to your Claude Desktop MCP configuration file (~/Library/Application Support/Claude/claude_desktop_config.json on macOS):

{
  "mcpServers": {
    "prerak-crm": {
      "command": "pnpm",
      "args": [
        "--dir",
        "/path/to/prerak2.0",
        "--filter",
        "@prerak/mcp-server",
        "start"
      ],
      "env": {
        "CRM_MCP_BASE_URL": "http://localhost:3000",
        "CRM_MCP_AUTH_MODE": "api_key",
        "CRM_MCP_API_KEY": "prk_your_workspace_key"
      }
    }
  }
}

Replace /path/to/prerak2.0 with your actual project directory and prk_your_workspace_key with your API key.

ChatGPT Integration

For ChatGPT integration, refer to the configuration and setup guide in the gpt-store/ directory. This includes custom GPT configuration and instructions for connecting to your Prerak workspace.

Example Usage

Create and Update a Contact

User: "Create a new contact for Sarah Johnson with email [email protected]"

Claude will:
1. Call contacts_create with name: "Sarah Johnson", email: "[email protected]"
2. Return the created contact with ID

User: "Add her phone number 555-1234"

Claude will:
1. Call contacts_update_properties with the contact ID and phone: "555-1234"
2. Confirm the update

Find Deals and Move Them Forward

User: "Show me all deals in the negotiation stage"

Claude will:
1. Call deals_list with stage filter "negotiation"
2. Display matching deals with values and owners

User: "Move the ACME deal to closing"

Claude will:
1. Call deals_update with the deal ID and new stage: "closing"
2. Confirm the stage change

Run an AI Agent

User: "Run the pipeline forecast agent for Q2"

Claude will:
1. Call agent_list to find available agents
2. Call agent_run with agent ID and parameters
3. Poll agent_status to track execution
4. Return results when complete

Send Email with Safety

User: "Send an email to [email protected] about the proposal"

Claude will:
1. Call emails_send with draft details
2. Request confirmation before sending (prevents accidental sends)
3. Execute when user confirms
4. Return confirmation with email ID

Analyze a Meeting

User: "What was discussed in the ACME meeting on Monday?"

Claude will:
1. Call meetings_list to find the meeting
2. Call meetings_transcript to retrieve the transcript
3. Call meetings_summary to generate an AI summary
4. Present findings to the user

Configuration Files

• .env.example - Example environment variables
• src/index.js - Server entry point
• src/tools/ - Individual tool implementations
• gpt-store/ - ChatGPT GPT configuration files

Support

For issues, questions, or feature requests, visit the Prerak documentation or contact support through the Prerak dashboard.