LLL

Ordinary languages let AI generate fast, then leave humans to clean up the mess. LLL pushes more discipline into the language itself: stronger structure, required specs, paired scenarios, behavioral verification, and visible pressure against quality drift as a project grows.

This extension brings LLL into VS Code.

Who It Is For

For technical people waiting for an AI-native language

LLL is a real language-and-compiler approach, not just a prompt style. It keeps familiar TypeScript syntax, then makes it stricter in the places where AI usually gets sloppy.

For ambitious builders who are not deeply technical

LLL is for people who want to build serious software with AI without becoming the permanent QA department. The goal is more control through named units, specs, and scenarios, not less involvement.

What This Extension Gives You

• create projects from bundled templates
• bootstrap a local workspace with bundled compiler assets
• work from a dedicated LLL sidebar inside VS Code
• use chat-driven generation and editing with compile/test feedback in one place
• choose providers and models, including OpenAI and OpenRouter support
• get guided prerequisite checks for local runtime setup

Why LLL Feels Different

• the compiler enforces more structure instead of relying on soft best practices
• @Spec and scenario-driven testing are first-class, not optional decoration
• behavioral verification is part of the workflow, not an afterthought
• coverage debt is surfaced so quality decay does not stay invisible

The point is not "magic" or "bug-free." The point is tighter control over AI-generated software as it stops being a toy project and starts becoming a real system.

Requirements

• VS Code
• Node.js

The extension handles the rest of the local bootstrap flow, including workspace dependency installation and bundled lllts compiler assets where applicable.

Learn More

• Website: https://lllts.dev
• Documentation: https://lllts.dev/documentation/introduction
• Repository: https://github.com/mathemaudio/lllts