HubSpot MCP Server

HubSpot MCP Server

Overview

This project implements a Model Context Protocol (MCP) server that integrates with HubSpot CRM and now includes a centralized shared space feature. The server enables AI models to interact with HubSpot data and operations through a standardized interface while allowing teams to share key conversation summaries in real time.

For more information about the Model Context Protocol and how it works, see Anthropic's MCP documentation.

Components

Resources

No resources are implemented. Ultimately, the system is driven by the tools.

Example Prompts

Create HubSpot contacts by copying from a LinkedIn profile webpage:

Get latest activities for your company:

Shared Space Example:

After a meeting, an AI command box prompts: "Kya aap is conversation summary ko company ke shared space mein share karna chahenge? (Yes/No)" On confirmation, the conversation summary is stored centrally and accessible to all authorized managers in your company.

Tools

The server offers several tools for managing HubSpot objects along with new shared space tools:

Contact Management Tools

• hubspot_create_contact
  • Create a new contact in HubSpot (with duplicate checking).
  • Inputs:
    • firstname (string)
    • lastname (string)
    • email (string, optional)
    • properties (dict, optional; e.g., {"phone": "123456789", "company": "HubSpot"})

Company Management Tools

• hubspot_create_company
  • Create a new company in HubSpot (with duplicate checking).
  • Inputs:
    • name (string)
    • properties (dict, optional)
• hubspot_get_company_activity
  • Retrieve the activity history for a specific company.
  • Input:
    • company_id (string)

Engagement Tools

• hubspot_get_recent_engagements
  • Retrieve recent engagement activities.
  • Inputs:
    • days (integer, default 7)
    • limit (integer, default 50)
• hubspot_get_recent_companies
  • Retrieve most recently active companies.
  • Input:
    • limit (integer, default 10)
• hubspot_get_recent_contacts
  • Retrieve most recently active contacts.
  • Input:
    • limit (integer, default 10)

Shared Space Tools

• create_shared_summary
  • Create a new conversation summary in the shared space.
  • Inputs:
    • user_id (string, the user’s email/ID)
    • company_id (string)
    • summary (string)
• get_shared_summaries
  • Retrieve all conversation summaries for a given company.
  • Inputs:
    • user_id (string)
    • company_id (string)

These shared space tools allow authorized managers to share meeting summaries, enabling real-time team collaboration and centralized access to key insights.

Setup

Installing via Smithery

To install buryhuang/mcp-hubspot for Claude Desktop automatically via Smithery:

Prerequisites

You'll need a HubSpot access token. Obtain one by creating a private app in your HubSpot account (see HubSpot Private Apps Guide). Keep your token secure.

Docker Installation

Option 1: Pull from Docker Hub

Option 2: Build Locally

Run the container:

Or pass the token as an argument:

Cross-Platform Publishing

Create a new builder instance:

Build and push for multiple platforms:

Verify the image:

Usage with Claude Desktop

Installing via Smithery

Docker Usage

Or using command-line argument:

Development

To set up the development environment:

Shared Space Integration

The shared space functionality is implemented in a separate module (shared_space.py). This module:

• Loads dynamic user roles (from an external JSON file or database).
• Implements CRUD endpoints (create_summary, get_summaries, update_summary, and delete_summary) for conversation summaries.
• Triggers a real-time notification placeholder (which you can later integrate with SSE or WebSockets).

The main server.py file imports these functions and registers them as new MCP tools (create_shared_summary and get_shared_summaries).

License

This project is licensed under the MIT License.

This updated README now reflects the additional shared space functionality using HubSpot MCP integration and explains how the system is structured and used.