HubSpot MCP Server
@baryhuang
About HubSpot MCP Server
A Model Context Protocol (MCP) server that enables AI assistants to interact with HubSpot CRM data, providing built-in vector storage and caching mechanisms help overcome HubSpot API limitations while improving response times.
Config
Add this server to your MCP-compatible client using the configuration below.
{
"mcpServers": {
"mcp-hubspot": {
"command": "npx",
"args": [
"-y",
"@smithery/cli@latest",
"install",
"mcp-hubspot",
"--client",
"claude"
]
}
}
}Tools
7Create contacts with duplicate prevention
Create companies with duplicate prevention
Retrieve activity for specific companies
Retrieve most recently active companies
Retrieve most recently active contacts
Retrieve recent conversation threads with messages
Semantic search across previously retrieved HubSpot data
Overview
What is HubSpot MCP Server?
An MCP server that connects AI assistants (like Claude) to HubSpot CRM data. It provides tools for managing contacts, companies, and engagement records, with built-in vector storage (FAISS) and caching to overcome HubSpot API limits and improve response times.
How to use HubSpot MCP Server?
Install via Smithery (npx -y @smithery/cli@latest install mcp-hubspot --client claude) or pull the Docker image (docker run -e HUBSPOT_ACCESS_TOKEN=your_token buryhuang/mcp-hubspot:latest). Configure the Docker container in your MCP client (e.g., Claude Desktop) with the HUBSPOT_ACCESS_TOKEN environment variable and an optional volume for persistent storage.
Key features of HubSpot MCP Server
- Vector storage using FAISS for semantic search across HubSpot data
- Thread-level indexing for precise retrieval of conversation records
- Embedding caching with SentenceTransformer for faster responses
- Persistent storage in a configurable directory between sessions
- Duplicate prevention when creating contacts and companies
- Multi-platform Docker images (linux/amd64, linux/arm64)
Use cases of HubSpot MCP Server
- Create HubSpot contacts and companies from a LinkedIn profile paste
- Retrieve recent pipeline activity and engagement updates
- Semantic search over previously fetched CRM data for context-aware queries
FAQ from HubSpot MCP Server
What tools does HubSpot MCP Server provide?
It provides seven tools: hubspot_create_contact, hubspot_create_company, hubspot_get_company_activity, hubspot_get_active_companies, hubspot_get_active_contacts, hubspot_get_recent_conversations, and hubspot_search_data.
What HubSpot permissions are required?
A HubSpot access token with scopes crm.objects.contacts (read/write), crm.objects.companies (read/write), and sales-email-read.
How do I deploy HubSpot MCP Server?
You can use Docker directly with docker run or install via Smitherly. For manual setup, run the Docker container and pass the token via HUBSPOT_ACCESS_TOKEN. Optionally mount a volume at /storage for persistent data.
Does HubSpot MCP Server store data locally?
Yes, it stores conversation threads and embeddings in a configurable storage directory (/storage by default). Data persists between sessions when the volume is mounted.
What is the vector storage used for?
The FAISS vector store enables semantic search across previously retrieved HubSpot data, allowing AI assistants to recall past CRM interactions without re-fetching from the API.
Frequently asked questions
What tools does HubSpot MCP Server provide?
It provides seven tools: `hubspot_create_contact`, `hubspot_create_company`, `hubspot_get_company_activity`, `hubspot_get_active_companies`, `hubspot_get_active_contacts`, `hubspot_get_recent_conversations`, and `hubspot_search_data`.
What HubSpot permissions are required?
A HubSpot access token with scopes `crm.objects.contacts` (read/write), `crm.objects.companies` (read/write), and `sales-email-read`.
How do I deploy HubSpot MCP Server?
You can use Docker directly with `docker run` or install via Smitherly. For manual setup, run the Docker container and pass the token via `HUBSPOT_ACCESS_TOKEN`. Optionally mount a volume at `/storage` for persistent data.
Does HubSpot MCP Server store data locally?
Yes, it stores conversation threads and embeddings in a configurable storage directory (`/storage` by default). Data persists between sessions when the volume is mounted.
What is the vector storage used for?
The FAISS vector store enables semantic search across previously retrieved HubSpot data, allowing AI assistants to recall past CRM interactions without re-fetching from the API.
Basic information
More Data & Analytics MCP servers
Jitsu
JitsuJitsu is an open-source Customer Data Platform — an MIT-licensed alternative to Segment. This MCP server gives your AI assistant direct control over a Jitsu workspace, so you can run and debug data pipelines in conversat
CryptoStruct Market Data
cryptostructRemote MCP server for historical crypto & prediction-market data: search ~500K instruments, live market stats (OHLC, turnover, spreads, depth, slippage), free sample files, and tick-data. Keyless for catalog & stats; opt
Context.dev
Yahia BakourContext.dev is the web-data infrastructure for AI products and agents, one API that turns the web into clean, structured, LLM-ready data. MCP available at https://mcp.context.dev.
Opinly
Track competitors, rankings, and how AI answers rank you — then write and publish the fix.
Bounce Watch - Company Signal Intelligence
bouncewatchBounce Watch tells you what just changed at a company, and when. Who raised money. Who hired a senior person. Who opened an office, won a customer, or announced a partnership. Every event carries its own date, so you ca
Comments