Into the Breach

What it is. A compact turn-based tactics game on an 8×8 grid. Enemies telegraph exactly what they’ll do next turn; you have three mechs and a fistful of moves to shove, block, and reposition so the incoming attacks land on nothing — or on each other.

Player fantasy / why it’s fun. I am a puzzle-solver defusing the future. The pull is perfect information: nothing is hidden, nothing is random-to-your-face, so every loss is your miscalculation and every clean turn feels like a proof you found. It’s chess where the opponent shows their move and dares you to invalidate it.

Design DNA

Invert the fog. Where most tactics hide enemy intent behind hit-chances and initiative, Into the Breach shows everything — targets, damage, order — and makes the game about editing that plan. Your verbs aren’t mostly damage; they’re displacement: push an enemy off its target, into another enemy, into a hazard. The threat is real, the answer is spatial, and the board is small enough to solve fully in your head.

Perfect information changes what “skill” means. In a hidden-info tactics game, skill is partly managing variance — hedging against the roll. Here there is no roll to hedge; skill is pure spatial reasoning, and a loss is unambiguously a puzzle you solved wrong. That honesty is a design constraint with teeth: every telegraph you show must resolve exactly as shown, or the whole contract collapses. The reward is that hard turns feel solvable rather than unlucky — the player always believes a perfect answer exists, and usually it does.

Constraint is the engine: tiny grid, few units, few moves, one objective — protect the buildings. Depth comes from interactions, not scale. Because you win by prevention rather than elimination, the state space stays small enough to reason about exhaustively, which is exactly why the puzzle lands.

Load-bearing structures

Structure	Why it works
Telegraphs as directions, not target tiles	Store “this bug attacks this way,” so pushing it rewrites the outcome. This is the single most important implementation choice. FUN.md truth. `system-telegraphs`.
Perfect information, perfect honesty	What’s shown is exactly what resolves — no hidden rolls. Fairness is structural, so hard turns feel solvable.
Displacement as the primary verb	Push/pull/block beats raw damage — you win by redirecting threats onto each other or into hazards. `system-combat-model` / `system-counter-systems`.
Tiny bounded board	8×8 keeps the state space small enough to fully reason about — the puzzle is tractable by design.
Protect-the-objective win condition	You don’t need to kill everything; you need the buildings to survive. Reframes “win” as damage-prevention. `system-encounter-design`.
Both proofs per scenario	A winning line AND a losing do-nothing must exist — a threat a null turn survives isn’t a threat. FUN.md law 4.
Run structure with persistent pilots	Light meta (pilot XP, mech unlocks) wraps the tactical puzzles into a campaign. `system-meta-progression`.

What to steal

Show the enemy’s move, then make the game about editing it. The whole design flips from prediction to manipulation. This is the reusable core.
Store telegraphs as directions on units, not fixed target tiles — so a push changes the outcome. (FUN.md §12; this is the load-bearing data choice.)
Make displacement a first-class verb. Redirecting a threat is more interesting than out-damaging it, and it turns enemies into weapons.
Keep the board tiny and the objective defensive. Solvability comes from bounded scale + “protect,” not “eliminate.”
Ship both proofs per encounter: a winning line and a losing null (FUN.md law 4); prove each mechanic — push, bump, redirect — in isolation. A 1-ply greedy bot over structuredCloned state is a real baseline defender and a completability check in one.
Reframe the win condition as prevention. “Protect the objective” instead of “kill everything” keeps the board small, makes displacement matter more than damage, and turns every enemy into a tool rather than just a target.

What’s just theme (drop it)

Kaiju / mech / time-travel fiction. Cosmetic; the puzzle is theme-blind.
The specific squads. Squad = a coherent verb-set with an identity is structural; “Rift Walkers vs Rusting Hulks” is flavour — system-faction-asymmetry.
The grid being 8×8. Small enough to fully reason about is the rule; the exact number is tuning.
Sci-fi VFX. The threat only needs to read its direction and damage clearly — pattern-readability.

Composes into

genre-tactics — its canonical anchor; the rewrite-the-future loop lives there.
system-telegraphs — the exemplar of intent-as-direction.
system-combat-model — displacement-first resolution.
system-counter-systems — turning threats against each other.
system-encounter-design — protect-the-objective scenario design.

Twist seams

Into the Breach but you play the swarm (perspective) — you set the telegraphs; a defender edits them; asymmetric two-side design. Feeds system-coop-and-competition / system-faction-asymmetry.
Into the Breach but real-time with a freeze verb (mechanic-swap) — the “shown future” becomes a pausable prediction line you scrub and rewrite. Bends the turn structure while keeping perfect information.
Into the Breach but the objective is to cause the chain reaction (constraint) — Rube-Goldberg tactics: you can’t attack directly, only nudge, so every kill is a redirect. Sharpens displacement into the only verb.
Into the Breach but tonal — you’re a crossing guard, not a soldier (tonal) — the telegraphed threats become traffic, weather, or a panicking crowd; you displace hazards away from people rather than bugs off buildings. Same perfect-information displacement puzzle, cozy register. Pairs with pattern-readability.