MCP Server + Local LLM (Ollama)

Run a local LLM with custom MCP servers to enable tool usage (files, APIs, automation) directly from your model.

This setup connects:

• A local LLM (Ollama)
• A custom MCP server

So your model can:

• Access external tools
• Read/write files
• Run custom automation workflows

MCP Server → exposes tools (functions)

MCP Client → connects those tools to the LLM

Ollama → runs the local model

The LLM can then call tools dynamically during responses.

Steps to set up your own LLM with Custom MCP Server -

1.) Install dependencies

pip install mcp-client-for-ollama

pip install mcp openpyxl pandas xlwings

2.) Install Model

ollama pull qwen2.5:7b

3.) Run MCP client

python -m mcp_client_for_ollama --servers-json "your path\servers.json"

Example Use Cases -

• Query Excel
• Generate reports
• Automate workflows