Stockfish
Objective engine analysis.
“What is objectively best?”
PlyMove
Chess intelligence built for teaching.
Give PlyMove a chess game or position and a learner Elo. It identifies what is worth explaining and returns learner-appropriate insights and variations.
Three questions. One pedagogical layer.
PlyMove does not replace Stockfish or human-move models. It combines them—with learned annotation priors and learner-relative scoring—to decide what to teach.
Human-move modelling
Plausible alternatives at a strength band.
“What is a player likely to consider?”
PlyMove
Learner-specific teaching decisions.
“What should this learner be taught?”
One concrete example
Homepage copy stays learner-facing. Expand for a sample developer payload.
Input
FEN: … (middlegame tactics)
target_elo: 1600Why this position matters
Qc4 looks natural, but it allows Qe7#.Qxe3 keeps the game going.
For a 1600-rated learner, this is worth highlighting because the losing move is human-plausible and the tactical punishment is immediate.
Developer output (illustrative)
{
"schema": "plymove/explain-position-v1",
"target_elo": 1600,
"teaching_reason": "critical_position",
"explanation": "Qc4 looks natural, but it allows Qe7#. Qxe3 keeps the game going."
}Capabilities
Explain a game
PGN + learner Elo → ranked teaching moments → annotated PGN / JSON
Explain a position
FEN + learner Elo → learner-specific explanation
Find teaching moments
Game → positions most worth explaining
Compare moves
Position + candidate moves → learner-relative comparison
Same intelligence, multiple surfaces
REST, MCP, and agent-paid calls share the Core analysis stack.
- REST APIProgrammatic access with API keys for products and backends.
- MCPRemote tools for AI agents and MCP-compatible clients.
- Agent-nativePaid HTTP calls where x402-compatible payment is enabled.
- Batch / embeddedVolume annotation and OEM integrations.
Built for AI agents
PlyMove can be called as a specialist chess capability by AI agents—without turning your agent into a chess engine.
“Explain this game for a 1450-rated learner.”
↓
AI agent calls PlyMove (MCP or REST)
↓
Structured teaching analysis returned
Surfaces: remote MCP, REST API, and x402-compatible agent payment where available.Agent payment details →
Embed learner-adaptive explanations
Add learner-adaptive chess explanations to your existing product without building the chess-teaching stack yourself.
Outputs include JSON, annotated PGN, teaching moments, variations, and pedagogical metadata.