See the tactic before you play it.
Chess games reward long plans, but many decisive moments are short and concrete: one move mates, one move forks two valuable pieces, one capture wins material, or one check forces the position into a narrow path. This specialist practice turns those moments into a repeatable local-first training loop instead of asking you to play a whole game every time you want ten minutes of tactical work.
Why this is separate from a normal chess game
A full game tests opening choices, calculation, time management, positional judgment and endgame technique all at once. That is useful, but it makes it difficult to isolate one tactical habit. The Tactics Academy keeps the existing complete Chess game intact and adds a second practice surface built for recognition, first-attempt accuracy and review.
Every V0.153 position is synthetic. The bank is not copied from a commercial puzzle feed, book, tournament archive or another training site. Positions are generated from deterministic legal playouts beginning at the normal starting position, then frozen only when the existing Chess legal-move engine can verify the target condition. The verifier re-parses every position, reconstructs its legal moves and confirms that the stored answer is legal and that it is the only move satisfying that puzzle's stated motif rule.
This matters because a tactics trainer should not grade an answer from a loose diagram or an unverified hand-written assumption. If the prompt says “find the only checkmate in one,” the engine checks every legal move for checkmate. If the prompt says “find the only fork,” the verifier checks the moved piece's attacks after every legal candidate. Winning Material requires a concrete favorable capture threshold, and Forcing Check requires the unique non-mating checking move. The goal is a transparent training contract, not a mysterious hidden engine score.
Four motif tracks
The bank is balanced at 24 frozen positions per motif, with both White-to-move and Black-to-move positions represented.
Mate in One
Scan checks, coverage and king escapes until you can identify the one legal move that ends the game immediately. The important habit is verification: a move is not mate merely because it gives check.
Fork
Find a single move that creates the required double attack against valuable enemy targets. The track includes positions where the tactical piece and target geometry are not presented in a textbook-clean arrangement.
Winning Material
Look for the capture that creates a large immediate material swing. This is deliberately concrete: the target must be valuable enough relative to the attacker for the verifier to classify the move as the tactic.
Forcing Check
Practice locating the only non-mating check in the position. This is a forcing-move scan rather than an “engine best move” claim, so the prompt and the grading rule stay aligned.
Study, Drill, then Mastery
Each motif uses the same visible three-step path. Nothing is artificially locked, so you can jump directly to the harder check while still seeing what has and has not been cleared.
Study explains the recognition target and records only that you reviewed the concept. Drill 12 draws twelve unique positions from that motif and clears at ten or more first-attempt correct answers. Mastery 16 uses sixteen unique positions and requires at least fourteen first-attempt correct answers. Across four motifs that creates twelve motif checkpoints. A thirteenth Mixed Mastery checkpoint draws exactly five positions from each motif for a twenty-puzzle test and clears at seventeen or more first-attempt correct.
First-attempt scoring is intentional. If you choose a legal but incorrect move, the puzzle does not disappear; you keep working until you find the target tactic. The session score still remembers that the first try was wrong. This separates learning from score inflation: you can finish the position and see the explanation, while the mastery threshold still reflects whether you recognized the pattern before feedback.
Weak Tactics means real misses only
The review queue is based on what actually happened on the board.
A puzzle enters Weak Tactics only after a genuine first miss or after you ask for a hint. Merely existing in the bank does not create weakness data, and unseen positions are never counted as errors. When you later solve that same puzzle correctly, it is removed from the active weak queue. Recommendation logic then looks for an incomplete motif with saved misses before falling back to the normal progression order.
This keeps review interpretable. If your Fork track has several saved misses, the Academy can point you back there without pretending that your Mate or Material track is weak. The review queue is local to this browser and does not create a global chess rating, public profile or leaderboard.
Sprint 60 is fluency, not mastery
Fast pattern recognition is useful, but speed should not silently replace accurate training.
Sprint 60 draws from the frozen mixed bank and counts how many positions you solve before sixty seconds expire. It keeps a local personal best, but it does not clear Study, Drill, Mastery or Mixed Mastery checkpoints. That separation matters because a fast run rewards recognition speed while the checkpoint path rewards first-attempt accuracy over a fixed number of puzzles.
Use Sprint after you already understand the motif or as a short transfer test. If your weak queue grows during a speed run, return to untimed Review and solve those positions carefully.
How to scan a tactical position
A consistent search order is more useful than guessing from the most dramatic-looking piece.
Start with checks. A check forces a response and can expose mating nets, king geometry and overloaded defenders. Then look at captures, especially captures where a lower-value piece can take a higher-value one. After that, inspect attacks that hit two valuable targets at once. The Academy's four tracks deliberately overlap with this practical scan order: Mate and Forcing Check teach checks, Winning Material trains concrete captures, and Fork trains double attacks.
Do not move only because a square “looks active.” Before committing, confirm that the move is legal, that your own king remains safe and that the resulting geometry matches the motif in the prompt. The board accepts source and destination clicks only after the same underlying ChessCore legality filter used by the classic game.
What this trainer does not claim
V0.153 is deliberately narrower than a full commercial chess training service.
The 96-position bank is not a replacement for hundreds of thousands of rated tactics, opening preparation, endgame tablebases, cloud engine analysis or coaching. It also does not assign an Elo-like tactics rating. The product is designed as a bounded, transparent local-first specialist inside DailyBrainArc: enough positions to build a real repeatable habit, exact motif rules that can be independently checked, weak-puzzle review, a visible mastery path and a speed transfer mode.
The existing Chess game remains available for full legal games against browser-local AI, including castling, en passant, promotion, checkmate and draw rules. Tactics Academy is an additional practice layer, not a rewrite of that game.
Why a synthetic bank is useful here
Real-game archives are valuable, but they are not the only way to build a trustworthy beginner-to-intermediate tactics loop.
A synthetic position can be useful when its construction and grading rules are explicit. In this project, the generator starts from the normal chess opening position and makes only legal moves through the same rules engine used by the playable Chess game. Candidate positions are accepted only when one legal move satisfies the exact motif rule. That means the bank can be regenerated and audited without depending on a remote service, a private puzzle rating, a copyrighted commentary layer or an opaque third-party API.
The bank also keeps the training target stable. A Mate puzzle is not silently reclassified as “best move” because an engine prefers a different continuation. A Forcing Check puzzle asks for the unique non-mating check, not the move with the highest hidden evaluation number. This narrow contract is especially useful for pattern recognition because the learner knows exactly what visual relationship to search for, and the verifier can test that same relationship independently.
That does not make synthetic positions universally better than master-game tactics. It makes them appropriate for this bounded local-first product: the positions are legal, reproducible, balanced across both sides to move, and tied to transparent mastery rules. A future expansion can add deeper multi-move calculation only when that additional authority can be verified just as clearly.
Start with one motif
If you are unsure where to begin, use Mate in One first. It has the clearest pass/fail condition and teaches the habit of checking every legal reply before calling a tactic finished.