A Model Context Protocol (MCP) server for searching HubSpot content using AIMS GenAI search capabilities. This project provides a service that can be integrated with AI assistants to fetch relevant content from HubSpot when answering user queries.
- MCP server implementation for AI assistant integration
- HubSpot content search API integration
- Support for both stdio (Claude) and SSE transport methods
- Express server for HTTP connections
- Node.js >= 22.0.0
- npm
Clone the repository and install dependencies:
git clone <repository-url>
cd aims-genai-search-mcp
npm install
The application uses environment variables for configuration. You can create a .env
file in the project root with the following options:
APP_VERSION=0.1.0
NODE_ENV=development
PORT=7878
TRANSPORT=stdio
Available transport methods:
-
stdio
: For command-line integration (default) -
sse
: For web integration
For production:
npm start
For development with auto-reloading:
npm run dev
The server exposes the following MCP tool:
-
search-hubspot-context
: Search content in HubSpot- Parameters:
query
(string) - The search query
- Parameters:
Build the TypeScript code:
npm run build
Watch for changes during development:
npm run build:watch
Inspect MCP server:
npm run inspect
# Build the image
docker buildx build -t hubspot/genai-search-mcp .
# Run the container
docker run -i --rm --name genai-search-mcp hubspot/genai-search-mcp
If you have the current repository cloned in your local machine, you can configure Claude Desktop to use it by running building the project and then adding the following configuration:
Alternatively you can use npx
:
{
"mcpServers": {
// ...
"genai-search": {
"command": "npx",
"args": ["@hs-web-team/aims-genai-search-mcp"]
}
}
}
Or Docker:
{
"mcpServers": {
// ...
"genai-search": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"--name",
"genai-search-mcp",
"hubspot/genai-search-mcp"
]
}
}
}