Chess Branching is a tool for building and studying opening repertoires. Its one idea: draw the variations as a diagram of connected moves, rather than burying them in brackets inside a paragraph of notation.
Open the boardChess notation was designed for printing a single game on paper. A repertoire is not a single game — it is a tree. Every time your opponent has a real choice, the line splits, and each of those branches splits again. Written in the standard way, that tree gets flattened into one long sentence:
Everything is there, but the shape is gone. You cannot tell at a glance how many replies you have prepared for, where the tree is thin, or which line is your main one. Nested brackets three levels deep are hard to read on paper and worse on a screen.
Chess Branching draws that identical repertoire as a graph. The main line runs straight across; each alternative opens its own row underneath, connected by an arrow:
Now the structure is obvious. Four replies to 1…e5, one of which is your main
line and three of which are short. The branch colours in the diagram match the arrows on
the board, so the moves you have prepared from the current position are visible without
clicking anything.
There is no “add variation” button. Click any move in the diagram to go back to it, play a different move on the board, and a new branch opens. Replay a move that is already in the tree and nothing is duplicated — you simply travel down that branch.
Every move can carry a note: the plan, the trap, the reason you chose this line. Moves that have a note are marked in the diagram, so you can see where your thinking is written down and where it isn’t.
A study does not have to begin from the initial array. Set up a position piece by piece, or paste a FEN, and branch out from there. Middlegame plans, endgame technique and tactical positions all fork the same way an opening does.
The bar beside the board shows who is better and by how much, the way it does on a broadcast, and the panel beside it lists the engine's three best continuations. Point at a move in one of them and the board shows that position; click the line and its first move joins your tree as a branch.
Stockfish runs inside your own browser, so nothing is sent to a server. Every position is searched to a fixed depth, which is why the same line always reads the same number.
Work is organised into studies — one for each opening, colour, or idea. You can fork a new study from any move, which copies everything below it into a fresh tree. That is useful when a sideline grows big enough to deserve its own file.
Import any PGN with variations and comments and it becomes a diagram. Export it back and you get standard PGN that opens in any other chess program. Nothing you build here is locked in.
You can start without an account, and then studies are saved in your own browser and never sent anywhere. Nothing to sign up for, nothing of yours on a server — but they do not follow you to another browser or device, and clearing your browsing data removes them. Use export PGN to keep a copy of anything you care about.
Making an account keeps them on the server instead, so the same repertoire is there on a phone and a laptop and survives a cleared cache. A free account keeps three studies, the same as a browser does. What we store, and how to delete all of it yourself, is set out on the privacy page.
Three is enough for most people. If it is not, a subscription lifts the limit: $4 a month, or $36 a year — the same as $3 a month — plus local tax where it applies. Nothing is deleted if it ever lapses: studies past the free three are locked rather than removed, and subscribing again opens them untouched. The terms say the rest.
Analysis is by Stockfish, compiled to WebAssembly and served with this site. Stockfish is licensed under the GNU General Public License version 3.
Open the board