A developer shares their experience rebuilding their VS Code setup from scratch after years of extension bloat slowed it down. The key changes: removing unused language, visual, and AI extensions, keeping only Prettier for formatting, and switching from Continue to llama-vscode for local LLM integration with llama.cpp. The result is a leaner, faster editor that pairs local models for autocompletion, embedding, and chat without relying on cloud-based AI subscriptions.

3m read timeFrom xda-developers.com
Post cover image
Table of contents
My VS Code instance had far too many extensionsIntegrating local LLMs with VS Code increased my productivity multifold
7.9K Impressions