Ramify
Your games, turned into your chess curriculum.
Opening courses teach you a straight line: if they play this, you play that. A real opponent can play almost anything, and the moment they do, a memorised line is a script with nothing left to say. Ramify studies the games you actually played, finds where you actually went wrong, and teaches the pattern behind it, so you still have a plan when no one is reading from your script.
Free forever*Open source, MIT*No account*Nothing leaves your browser unless you ask it to
You know this feeling
You studied an opening line down to move nine. Your opponent played something else on move four.
Now you are improvising against someone who was never going to follow your notes, and in the moment you cannot tell whether your position is fine or already lost. Ten moves later you resign, or you win anyway and are not sure why. Either way, you did not learn anything you can use next time.
That is not a memory problem. What you memorised was a path through one specific game, not a way of understanding any game, and a path with no more branches is a dead end the moment someone steps off it.
How it works
One loop, run after every game, that turns what actually happened at the board into what you study next.
-
Play, or bring in a game
Play the built-in Stockfish opponent at a level that fits you, or bring in a game you already played on Lichess or Chess.com.
-
The engine finds your real mistakes
Stockfish 18 scores every position in the game, right in your browser, and flags exactly where the evaluation, and your win chance, dropped.
-
The coach explains the pattern, not just the move
Each mistake is tagged with the real chess reason: a hanging piece, a missed fork, an overloaded defender, a bad bishop, one of 23 recognised patterns, explained in plain English.
-
You drill it until it stops happening
The position you got wrong becomes a puzzle. Spaced repetition brings your worst patterns back exactly as often as you still need them, and lets go once you don't.
Then you play again, and the curriculum updates itself.
Three jobs, one loop
Learn, train and review are not three different products bolted together. They are three views of the same weakness profile.
-
Learn
Principles that survive deviation
When your opponent leaves known theory, which happens to almost everyone by move ten, Ramify does not go quiet. It names what changed on the board and which principle applies, and it can turn any pattern that keeps costing you into a short, plain-English lesson.
-
Train
Drills from your own blunders
Every mistake becomes a puzzle set at the exact position where you went wrong, not a generic puzzle about someone else's game. A weakness profile ranks what is actually costing you, and a Leitner-style schedule keeps bringing a pattern back until you have earned the right to stop seeing it.
-
Review
A real reason for every mistake
Stockfish 18 reviews every move: centipawn loss, win-probability swing, and a classification from best to blunder. Then the pattern engine explains why, using static exchange evaluation, forks, pins, skewers, overloaded defenders, pawn structure and king safety.
- Good
- Inaccuracy
- Blunder
See the whole position, not one line through it
Opening prep usually shows you one move at a time: the one your course expects. The explorer shows you the branch. From any position you reach, it lists every legal reply, names the ones that are recognised theory, and explains what each one is actually for.
Example, straight from the app's own opening data
After 1. e4 e5 2. Nf3 Nc6 3. Bb5 (Ruy Lopez, ECO C60), all 30 legal replies are known. Here are the named ones:
- 3...a6 Ruy Lopez: Morphy DefenseC70
- 3...Nf6 Ruy Lopez: Berlin DefenseC65
- 3...f5 Ruy Lopez: Schliemann DefenseC63
- 3...Bc5 Ruy Lopez: Classical VariationC64
- 3...Ba3 not in the book here, marked honestly rather than hidden
Pick any reply and the explorer explains itself: what it is called, how often it is actually played, and the plan behind it, for example "the main line, played in most games from here" or "a sideline, seen in about 3% of games from here". Run past the last named move and the explanation does not stop: it falls back to the same short list of principles a strong player reaches for once preparation runs out, who controls the centre, whose piece is doing the least work, are both kings safe, what does the structure promise or threaten.
Built from Lichess's public, CC0-licensed opening data, not memorised by you, just available to you the moment you need it.
Why this is not just another engine, or another opening trainer
Compared with generic engine analysis
Plenty of tools will tell you a move lost 0.8 of a pawn. Fewer will tell you why: a piece was overloaded, a bishop was bad, your king had one fewer defender than the position needed. A number is an evaluation. A reason is a lesson, and it is the reason that lets you recognise the same mistake somewhere else.
Compared with memorising opening lines
Other tools help you memorise more lines, more deeply. That works exactly until your opponent leaves the line, which below master level is usually before move fifteen. Depth of memorisation does not fix a shallow understanding of why the moves were good, and understanding is the part that has to do the work once you are on your own.
Ramify starts from your actual games, tags what actually goes wrong, and teaches the principle underneath it, the part that stays true no matter what your opponent plays next.
Runs on your machine. Costs nothing. Answers to no one.
Stockfish is compiled to WebAssembly and analyses every position on your machine; there is no server watching. There is no account to create and nothing to upload: your games and your weakness profile live in your browser's local storage.
The only network calls Ramify can make are ones you choose to trigger yourself: bringing in a game you already played on Lichess or Chess.com, or fetching an extra themed puzzle from Lichess's public API. Everything else, every evaluation, every pattern tag, every drill, every lesson, runs locally and stays there.
- No account, ever
- Free today and it will stay free
- MIT licensed, source hosted in the open
- Fork it or run your own copy any time
Frequently asked questions
Is it really free?
Yes. There is no tier, no paywall and no trial period. The hosted version costs nothing to use, and the source is MIT licensed if you would rather run your own copy.
Do I need an account?
No. There is no sign-up and no login anywhere in the app. Open it and start playing or reviewing.
Does it work offline?
Once the page and the engine have loaded, yes: playing, reviewing, tagging patterns and drilling all run locally with no further network calls. Loading the app for the first time, or after clearing your cache, still needs one connection, the same as any web page; after that, the chess itself does not need the internet.
What engine is it?
Stockfish 18, one of the strongest chess engines that exists, compiled to WebAssembly and run in a background thread in your browser. It is the same analysis a strong desktop engine would give you, not a lightweight imitation.
Can I import my Lichess and Chess.com games?
Yes. Bring in a game you have already played on either site and Ramify reviews it exactly like one you played inside the app, including naming the opening and generating drills from anything you got wrong.
Is my data sent anywhere?
Not unless you ask it to be. Your games and profile stay in your browser's local storage. The only outbound calls are ones you trigger yourself: bringing in a game from Lichess or Chess.com, or fetching an extra Lichess puzzle. If you self-host with your own LLM key for richer coaching text, the structured review data for that one request goes to the provider you configured, never anywhere else, and never by default.
How is this different from Chess.com's Game Review?
A game review gives you a strong per-move evaluation and a short comment on the position you just finished. Ramify does that too, then keeps going: it names the recurring reason a mistake keeps happening, not just in this game but across all of them, and turns it into a scheduled drill until it stops. It also has an answer for the moment theory runs out, which a single game's review was never trying to solve.