ChessMe

ChessMe is a personal chess trainer that turns your own games into analysis, coaching, recurring leak detection, and targeted drills.

It is built for players who want more than a generic engine review. Instead of only showing mistakes, ChessMe tries to answer:

• what kind of mistakes keep repeating
• where your thinking breaks down
• what to train next
• how to turn your own positions into a practical improvement loop

What It Does

• Open a real Chess.com profile by exact username
• Import completed games from Chess.com by username
• Upload PGN files as a fallback
• Run Stockfish-backed engine analysis on your games
• Detect recurring weaknesses such as tactical oversights, opening leaks, and endgame errors
• Create training cards from your own critical positions
• Add grounded AI coaching on top of engine facts
• Let you chat with a coach about a game, move, leak, or recent trend
• Save private notes tied to games, moves, openings, leaks, and coach answers
• Save local profile shortcuts and switch between active profiles quickly

Why It Is Different

Most chess review tools stop at evaluation swings.

ChessMe is designed around an improvement loop:

1. Import games
2. Analyze mistakes
3. Cluster repeated leaks
4. Review critical moments
5. Train on your own positions
6. Track whether your recent games are getting better

Core Features

• Dashboard control room for choosing a profile, syncing games, and launching analysis
• Dashboard with recent games, favorites, weakness clusters, and training status
• Game review page with board replay, critical moments, coach chat, and per-game notes
• Leak pages with examples from your own games, practical fix guidance, and related drills
• Coach Lab with blindspot map, recent trends, and style-focused AI coaching
• Training queue built from your own mistakes instead of generic puzzle sets
• Notes hub with search and contextual retrieval
• Global active-profile switcher in the header
• Light and dark theme support
• Mobile-friendly responsive layout

Stack

• Next.js App Router + TypeScript
• SQLite + Drizzle ORM
• chess.js for PGN/FEN handling
• local Stockfish with deterministic fallback coaching
• provider-based LLM layer with OpenAI first

Setup

npm install
cp .env.example .env.local
npm run dev

Open http://localhost:3000.

Configuration

• Chess.com import uses the public archives API
• Profile lookup uses an exact public Chess.com username
• AI provider, model, and token are managed in-app at /settings
• Core game analysis works without an AI token

How AI Works

• Without a token:
  • ChessMe still analyzes games with Stockfish-backed engine review
  • Coaching falls back to a deterministic local provider
• With a token:
  • You unlock deeper coach chat, leak explanations, and style reports
  • AI is used to explain engine-backed facts, not replace the engine

Public vs Private Data

ChessMe uses a hybrid storage model:

• Public/server data:
  • profiles
  • imported games
  • engine analysis
  • weakness patterns
  • training card templates
• Private/local browser data:
  • OpenAI token and AI settings
  • notes
  • favorites and saved profiles
  • coach chat history
  • private AI reviews, reports, and leak explanations
  • local training progress

This means the factual chess workspace can be hosted publicly, while your personal coaching layer stays on your device.

Privacy

• Public server data is stored in the app database under ./data
• Private personal data is stored in the browser on the current device
• Active profile selection is browser-local and not shared globally
• .env.local and runtime data are gitignored
• If AI credentials are missing or an AI call fails, the app falls back to deterministic summaries
• If Stockfish is unavailable, the app falls back to a simple material-based evaluator so the pipeline remains usable

Status

This project is currently optimized for a coach-first personal workflow, with public chess data on the server and private coaching data kept locally on-device.