The method
Five minutes to place you, a plan that rewrites itself after every session, lessons where you touch the board every thirty seconds. Nothing here is decided by guesswork.
FreeNo card requiredFive minutes
The sequence
Every step produces a number, and that number decides the next one.
Diagnosis
Three to eight adaptive positions, then a game against an engine calibrated to your presumed level. The verdict is measured: estimated level, strengths, biggest leak.
5 minand you know where you stand, within ±40
Plan
The module costing you the most points moves to the front, and the order is recomputed without you asking for anything. Every change is explained in one sentence.
3triggers for a recompute: module done, test, rating slip
Lesson
Demonstration, check, guided exercises, validation test: every notion is played. When you get stuck twice, the coach changes angle instead of repeating itself louder.
30 sof reading at most without touching the board
Exam
Ten unseen positions, two attempts each, no hints and no commentary. Missed positions come back in your reviews the next day.
70 %correct to move up to the next grade
The two rules
The LLM coaches, the engine plays. Neither does the other one’s job.
It keeps enough to tell you, three weeks later: "remember what we said about diagonals". That memory belongs to you — you read it and erase it from your settings, with no delay.
Every claim about a position comes from Stockfish, running in your browser. The coach computes nothing in its head: it asks, and if the engine contradicts it, the engine is right.
100 %of positions verified by the engine before publication
The diagnosis is free and asks for no card. At the end you have your estimated level and the first module of your plan.