home / mcp / leave manager mcp server
A self-hosted AI stack combining Ollama for running models, Open WebUI for user-friendly chat interaction, and MCP for centralized model management—offering full control, privacy, and flexibility without relying on the cloud.
Configuration
View docs{
"mcpServers": {
"ahmad-act-local-ai-with-ollama-open-webui-mcp-on-windows": {
"url": "http://localhost:8000/openapi.json"
}
}
}You run a self-hosted AI stack that combines Ollama for local language models, Open WebUI for a friendly chat interface, and MCP for centralized model management. This setup gives you full control over your AI tools, keeps data private, and makes it easy to manage and extend your MCP-powered services from a single place.
You access the Leave Manager MCP server from your MCP client or Open WebUI. The MCP server exposes its API via OpenAPI at the server base, and you integrate tools by adding them to Open WebUI so users can check leave balances, apply for leave, view histories, and personalize greetings.
Typical usage patterns include loading the Leave Manager tool in your MCP-enabled client, then using the available operations through the UI to request leave data, submit new leave requests, review past leave activity, and customize responses with a greeting. You can discover available endpoints through the OpenAPI interface exposed by the MCP server and then interact with those endpoints through your client’s tool integration flow.
# Prerequisites
# 1) Windows 10 or later
# 2) Docker Desktop for Windows installed and running
# Install Ollama (local model runtime)
# Follow the installer flow to completion. Ollama will run as a background service.
# Start Ollama server (if not already running)
ollama serve
# Verify Ollama is running
curl http://localhost:11434/api/tagsInstall and run the MCP-based Leave Manager alongside Open WebUI using Docker Compose. You will launch both the MCP server and the UI with a single command.
# Clone the project repository and navigate to the project directory
git clone https://example.com/Local-AI-with-Ollama-Open-WebUI-MCP-on-Windows.git
cd Local-AI-with-Ollama-Open-WebUI-MCP-on-Windows
# Build and start all services (MCP server + Open WebUI)
docker-compose up --buildConfiguration notes: The MCP server is exposed to the network on port 8000 for API access and to port 3000 for Open WebUI. The OpenAPI specification for the MCP server is available at the base API URL via the /openapi.json path. You can add MCP tools to Open WebUI by opening the UI, going to Settings > Tools, and registering the MCP endpoint at the root of the MCP server.
Security considerations: Run the MCP server behind your trusted network and limit access to the UI and API to authorized users. If you expose the MCP endpoints beyond your local network, enable authentication and review access logs regularly.
Troubleshooting: - Ollama not running: ensure the service is active and accessible at http://localhost:11434. Use ollama serve to start it if needed. - Docker issues: confirm Docker Desktop is running and you have sufficient disk space. - Model not found: verify the deepseek-r1 model is listed with ollama list. - Port conflicts: ensure ports 11434, 3000, and 8000 are free on your machine.
Query current leave balance for an employee to determine available days.
Submit a new leave request for specific dates and employee.
Retrieve historical leave records for an employee.
Return a greeting customized by user input to personalize interactions.