Elevate your chess game with the first extension that runs state-of-the-art neural models locally. Switch between raw Stockfish horsepower and anti-cheat safe human intuition.
ChessCompanion packs professional features into your browser tab overlays, executing client-side with absolute performance and privacy.
Choose between Stockfish 18 WASM for brute-force tactical precision and the Maia neural network model for human-like positional play. Perfect for studying theory or practicing against realistic opponents.
Calculates opponent candidates on their clock. Displays expectations as subtle dotted arrows, allowing you to prepare replies early.
Learn theory step-by-step. Integrates popular books (e.g., Queen's Gambit, Sicilian Najdorf) and seamlessly transitions to calculations once you leave book paths.
Condition the neural network weights on specific rating levels for both players. Simulate how a 2400 Grandmaster exploits a 1200 Intermediate, or how a Club player defends against high-pressure positions, mirroring real tournament psychology.
Understand the computational methodologies behind ChessCompanion's dual-engine interface.
Classical brute-force search computing millions of candidate trees to locate mathematically absolute moves. Ideal for tactical refutations and game analysis.
A 5M parameter neural net trained on actual human games. Maia evaluates positions using intuitive pattern matching rather than calculation, simulating human-like moves and blunders.
How ChessCompanion parses DOM board changes, processes evaluations, and handles runtime WebAssembly inside browser restrictions.
Attaches observers to active board elements. Converts square coordinates into standard FEN strings with instant turn-sensing hooks.
A background MV3 service worker that coordinates message queues between popup controls, active tabs, and engine sandboxes.
Bypasses extension security constraints. Spawns sandboxes with custom headers to load parallel calculation runtimes.
Loads computation runtimes client-side. Executes parallel calculation blocks to render immediate vectors on your tab board.
How ChessCompanion throttles search depth and tactical pruning to replicate realistic play tiers when Limit Strength is enabled.
ChessCompanion hooks directly into standard browser document trees, parsing elements without triggers or account flags.
Compatible with standard, custom, and beta board interfaces.
Native canvas overlay rendering for smooth, zero-latency vector guides.
We prioritize account safety. Unlike standard engines that calculate with mechanical rigidity, ChessCompanion is designed to emulate human signatures.
Maia is trained on games played by real humans of specific rating bands. It chooses moves based on human positional intuition rather than infinite tree searches, ensuring play patterns mimic human Elo tiers and blunder probability.
Your license key validation, chess calculations, and model execution run entirely client-side. The extension makes zero requests to external chess evaluation APIs, guaranteeing your play data remains isolated and completely private.
Answers to typical configuration, compliance, and engine capability questions.