Enables querying and managing knowledge from EKMS via MCP endpoints for search, recommendations, item retrieval, and feedback.
Configuration
View docs{
"mcpServers": {
"jihsin-ekms-mcp-server": {
"command": "node",
"args": [
"/path/to/ekms-mcp-server/src/index.js"
],
"env": {
"EKMS_API_KEY": "your-api-key",
"EKMS_API_URL": "http://localhost:3000"
}
}
}
}This MCP server enables Claude Code to query and interact with your Enterprise Knowledge Management System data. It provides a bridge that lets you search, fetch, and understand knowledge items, while also recording feedback and surfacing knowledge graphs to improve response quality.
Connect your MCP client to the EKMS MCP server to perform knowledge operations from Claude Code or the CLI. You can start the server locally and reference it in your client configuration, so you can search the knowledge base, get full knowledge items, list knowledge types, and view recommendations based on context.
Prerequisites you need before installing this MCP server:
- Node.js installed on your machine.
# 1) Install dependencies in the MCP server folder
cd ekms-mcp-server
npm install2) Create a local environment file based on the template and customize values for your EKMS API connection.
cp .env.example .env3) Edit the environment variables to point to your EKMS API and provide the API key.
# .env
EKMS_API_URL=http://localhost:3000
EKMS_API_KEY=your-api-key4) Start the server locally to verify it runs and listens for MCP connections.
npm start5) If you are configuring Claude Code, provide the MCP connection using the following example configuration.
{
"mcpServers": {
"ekms": {
"command": "node",
"args": ["/path/to/ekms-mcp-server/src/index.js"],
"env": {
"EKMS_API_URL": "http://localhost:3000",
"EKMS_API_KEY": "your-api-key"
}
}
}
}If you prefer using the Claude CLI, you can add the MCP endpoint with a direct command that runs the MCP server locally.
claude mcp add ekms node /path/to/ekms-mcp-server/src/index.jsEnvironment variables are used to connect to the EKMS API. Keep your API key secure and share it only with trusted clients. The MCP server exposes a set of capabilities to interact with knowledge data, including search, recommendations, item retrieval, and feedback recording.
Development workflow includes a dev mode for automatic reloading and tests to ensure functionality. You can run development and test commands from the project root.
Treat EKMS_API_KEY as a secret and rotate it periodically. Use network access controls to restrict who can connect to the EKMS API and the MCP server, and store credentials in a secure location appropriate for your environment.
Search the knowledge base with support for hybrid, semantic, and keyword search modes.
Recommend knowledge items based on context and customer characteristics.
Fetch the full content of a specific knowledge item.
List all knowledge types available in the system.
List all customers stored in the EKMS.
Retrieve the knowledge graph showing relationships between items.
Record feedback on knowledge usage to improve future results.