Caissa

A desktop chess workbench — a lighter ChessBase. It keeps your games as ordinary PGN files on your own disk, analyses them with a bundled Stockfish, builds and drills opening repertoires, imports from lichess, chess.com and ChessBase exports, and trains blindfold visualisation.

Everything runs locally. There is no account, no cloud library, and no telemetry. The network is used only when you ask for something that lives on it: a lichess import, an endgame tablebase lookup, a Wikipedia summary.

Start here: Getting started · coming from ChessBase? Importing games · want your lichess studies? Your lichess account

What it does

A database you own	Games live as PGN text in folders you can read, copy and back up. SQLite is only an index over them — delete it and your games are still there.
Analysis that explains itself	Live Stockfish with up to five coloured lines, full search and machine telemetry, and one-click annotation of a whole game into your PGN.
Openings, judged	Walk a collection’s tree move by move and see how it went for whoever played it — scores from their side, trends by year, and the lines that cost the most points.
Repertoires that drill	Import a lichess study or any PGN with variations; every line becomes a spaced-repetition drill you play blindfold.
Study material in place	Position-indexed history, your own pinned notes, PDFs beside the board, and Wikipedia background for the game you are reading.

Guide

• Getting started — install, run, and where your library lives

• The game database — collections, searching, sorting, deleting

• Importing games — PGN, archives, online accounts, ChessBase

• The opening explorer — walking a collection’s tree

• The analysis board — engine, telemetry, annotation, context

• Play the computer — eleven Stockfish levels, takebacks and saved games

• Repertoires — building lines and drilling them

• Your lichess account — studies, and games that import themselves

• Move sounds — the bundled sets and your own

• Blindfold training — the seven levels

• Workbench reference — every control, module by module

Reference

• HTTP API — every route the interface uses

• Data model — the library on disk and in SQLite

• Python modules — generated from the source

Development

• Architecture — how the pieces fit together

• Testing — the suites and what they cover

• Building the executable

• Releasing — the three platform builds, and how they are made

• Licensing and credits