Proposal Draft — Privacy-First Local AI Assistant for Joplin
Contributor: Syed Abrar Ahmed
GitHub: SYEDABRAR037 (SYED ABRAR AHMED) · GitHub
Introduction
Joplin users accumulate hundreds of notes over time but have no intelligent way to interact with them. Existing AI tools like ChatGPT require sending private note content to external servers — fundamentally against Joplin's privacy-first philosophy. This project builds a fully local AI assistant powered by Ollama, so users can ask natural language questions about their notes with zero data leaving their machine.
Project Goals
Local LLM integration layer in Node.js/TypeScript via Ollama's REST API
React/TypeScript AI chat panel in Joplin's desktop UI using the Plugin API
Smart context selection engine using BM25 + recency-weighted re-ranking
Settings UI for Ollama endpoint, model selection, and token budget
Jest unit tests for all modules written throughout the programme
Full user-facing documentation written incrementally
Why this idea?
This project sits at the exact intersection of Joplin's two 2026 priorities — AI/ML and Security/Privacy. Most AI note tools send data to the cloud. This one never does. I believe that is exactly what Joplin's community needs.
I am actively working on my first contribution to the codebase. Would love mentor feedback on this proposal — especially on the context selection approach and plugin architecture. Thank you!