HubSpot MCP Server
HubSpot CRM contacts, companies, and activities managed via AI-ready toolsets.
Data last scanned today · Reviewed today
Overview
This MCP server bridges your AI assistant directly with HubSpot, allowing for automated management of CRM objects. It includes specialized tools to create contacts and companies, retrieve activity logs, and fetch recent conversation threads directly from your HubSpot environment. Built with performance in mind, the implementation features integrated FAISS vector storage and caching mechanisms. These features enable semantic search capabilities over retrieved data, allowing the model to maintain context across CRM workflows and multi-step interactions.
Our verdict
This server carries a high-risk rating with a security score of 39/100, primarily due to a lack of read-only operational modes and missing OAuth support, which forces reliance on full-access API tokens. Users should exercise caution as the project is community-maintained rather than official and lacks recent updates, with the last commit occurring over 260 days ago. It is suitable for testing or personal environments where access control is tightly managed, but not recommended for production instances requiring stringent security boundaries.
- Integrated FAISS vector storage for improved semantic search capabilities.
- Zero direct dependencies minimize the software supply chain surface.
- Includes duplicate prevention logic for contact and company creation.
- Requires full write access tokens with no read-only mode.
- Lacks OAuth support for secure, scoped token management.
- Community-maintained project without official vendor backing.
- Significant inactivity with the last code commit 262 days ago.
Tools
| Tool | Description | Risk |
|---|---|---|
| hubspot_create_contact | Create contacts with duplicate prevention | medium |
| hubspot_create_company | Create companies with duplicate prevention | medium |
| hubspot_get_company_activity | Retrieve activity for specific companies | low |
| hubspot_get_active_companies | Retrieve most recently active companies | low |
| hubspot_get_active_contacts | Retrieve most recently active contacts | low |
| hubspot_get_recent_conversations | Retrieve recent conversation threads with messages | low |
| hubspot_search_data | Semantic search across previously retrieved HubSpot data | low |
Compatibility
| Client | Local | Docker | Remote | Read-only |
|---|---|---|---|---|
| Claude Desktop | ||||
| Cursor | ||||
| VS Code | ||||
| Windsurf | ||||
| ChatGPT |
Frequently asked questions
›What HubSpot scopes are required to use this server?
You must provide a HubSpot access token with crm.objects.contacts (read/write), crm.objects.companies (read/write), and sales-email-read scopes.
›Does the server support persistent storage?
Yes, you can configure persistent storage by mapping a local directory to the /storage path in your Docker container.
›How does the server handle HubSpot API limitations?
The server incorporates built-in vector storage and caching mechanisms to improve response times and mitigate API request limits.
›Is there a specific tool for finding information across retrieved data?
Yes, you can use the hubspot_search_data tool to perform semantic search across previously retrieved CRM information.
›Are there built-in features to prevent duplicate entries in HubSpot?
Yes, the hubspot_create_contact and hubspot_create_company tools are designed with built-in duplicate prevention.
Badge
Maintain this server? Add the live badge to your README.