Discovery Learning Engine (Antigravity Skill)

A custom skill for the Google Antigravity (AGY) system that transforms the AI into a strict Socratic tutor. Instead of lecturing, it guides you to learn technical subjects (like Data Structures & Algorithms) through a cycle of guided self-discovery: hitting a wall, hypothesizing a solution, and then revealing the true invention.

Features

• State Machine Guided Learning: Forces the user through a strict pedagogical loop (DISCOVER → STRUGGLE → HYPOTHESIS → REVEAL → IMPLEMENT → REFLECT → REPS).
• Token Efficient: Designed to keep LLM context windows small by maintaining a minimal progress.md state file and writing dense concept notes rather than re-reading long chat histories.
• Subject Agnostic: Originally designed for DSA, but adaptable to any technical subject where problems organically motivate the need for new tools.

Installation

Clone this repository into your local Antigravity skills directory:

git clone git@github.com:codebyvimal/discovery-learning-engine.git ~/.gemini/config/skills/discovery-learning-engine

Usage

Once installed, trigger the skill in a new Antigravity session by explicitly mentioning it:

• "I want to learn [Subject] using discovery learning."
• "Start a discovery learning session on [Topic]."

The agent will automatically initiate the setup interview and begin the discovery loop.