1. How It Works
When a student finishes a game on the platform, the position history is immediately run through Stockfish WebAssembly — the strongest open-source chess engine, running directly in the browser. Every move is evaluated for centipawn loss.
Moves with centipawn loss above a configurable threshold are flagged as blunders. Gemini AI then classifies the tactical theme of each blunder (hanging piece, missed fork, back-rank weakness, failed endgame conversion) and triggers an automated puzzle assignment matching that specific weakness.
2. Automated Puzzle Assignment
The moment a blunder is detected, the system queries the 6M+ puzzle database for positions that directly train the identified weakness. The student sees their assigned puzzles the next time they open their dashboard — before they've even had time to forget the game.
This closes the loop between mistake and remediation automatically. Coaches don't need to review every game to spot these patterns — the system surfaces them.
3. Coach View
From your coach dashboard, the Students section shows each student's recent game analysis summary — top blunder types, centipawn loss trend over time, and current assigned puzzle sets. You can drill into any individual game for the full move-by-move evaluation.