This repo contains the Astro blog and example code used in 🐳 Blog post: Teaching a local LLM to explore my notes with MCP .
It is a simple FastMCP server exposing tools and resources over a notes sandbox. Scripts and configs to run experiments with local models via Ollama are also shared.
Status: experimental / WIP.
Tested with: 16 GB VRAM GPU, Ollama, FastMCP, granite4:7b-a1b-h, llama3.2, qwen, gemma3-tools, mistral-nemo.