Lessons for my own (Godot) games

Spire's mechanics are some of the most transferable in modern indie design. Nothing here is locked to deckbuilders.

Steal: pick a dialectic, restate it fractally

Spire's whole game is risk vs reward at every scale. Every screen presents a tradeoff, and they're all the same shape. New players intuit the principle in 30 minutes because the game shows it to them six different ways.

For my own games: what's my one phrase? "Speed vs accuracy"? "Greed vs survival"? "Specialize vs adapt"? Whatever it is, every screen should be a restatement.

Steal: the Ascension staircase

A 20-step difficulty staircase, one new constraint per level, per character. Works for almost any genre with a "did you win?" loop. Hades does it (Heat). RoR2 does it (Eclipse). Could trivially work for an action game, a strategy game, a tower defense, a survival roguelite.

Way more durable than a "Hard / Hell" slider. Each level is a digestible new lesson; players self-pace.

Steal: enemy intent telegraphs

The single most-transferable Spire mechanic. If your enemies do something next turn / next interval, show it. Removes feel-bad RNG, replaces it with planning skill.

Works for turn-based combat, semi-real-time, even action games (Dark Souls already does this with attack windups). Sparklite even does a softer version with enemy wind-ups. Make the AI's next move visible.

Steal: power as variety, not stats

For replay-heavy games, meta-progression should unlock variety, not raise base power. The next run should be more interesting, not easier. New cards, new relics, new modifiers, new starter loadouts — yes. Bigger HP, bigger damage — no.

This is a strong default if you want runs to matter long-term. Hades does the opposite (and is great), but Hades is a very different shape — its meta-progression is deliberately a power ramp because the game wants you to eventually win. Spire's stance: keep the run's challenge constant; expand what you might encounter.

Steal: subtractive loadouts

Spire-style "your starting loadout is bad and you carve it down" is a fertile space and almost no Godot indie has tried it outside the deckbuilder genre. Could work for spell loadouts (start with a bloated kit, prune to a focused build), ship modules, ability bars, character rosters.

The lesson: thin decks > big decks. Make removing as rewarding as adding.

Steal: the map as a meta-decision layer

Adding a "node map" between fights — even a simple one — turns "string of random encounters" into "chosen sequence of risks." Huge perceived-agency upgrade for very little authoring cost.

Could overlay on top of an action game (Hades' isometric room choice is a softer version of the same idea). Could overlay on top of a level-based platformer. Anywhere the player would otherwise just be handed encounters in a fixed order.

Steal: card draft with skip

The "pick 1 of 3 or skip" pattern. Generalizes far past cards. Treat 'nothing' as a first-class option in any reward draft. It changes player decisions from "which is best" to "is any of these worth my limited slot."

Steal: bonus-with-drawback for run-defining loot

Almost every Spire boss relic has a drawback. This forces commitment. If your game has run-defining loot, give it costs. Uncosted upgrades flatten variety because every "great" run has the same shape.

Steal: daily seed = community for free

If your game has any RNG, a daily-seeded mode with a leaderboard costs almost nothing and creates community engagement that traditional patches can't. Wordle proved this beyond doubt. Make all your RNG seedable from day one.

Steal: anti-streak pity systems

Hidden counters that bias drops upward when the player has been unlucky. Smooths feel-bad runs without telling the player. Players don't notice the pity system existing; they just notice that the game rarely feels malicious.

This is genuinely worth doing for any game with random rewards.

Cautionary lesson: don't trust playtester preferences over data

Mega Crit's metric-driven patching is what kept Spire from drift over years of updates. Need a way to track something — pick rates, win rates, completion rates — even in a small game. Even a simple analytics layer is more valuable than another playtester opinion round.

Cautionary lesson: be ruthless about pruning

Mega Crit generated thousands of card ideas and ruthlessly pruned weak concepts. Most games ship too much. Spire shipped what survived testing.

For my own games: design more than I ship. Let half of what I prototype get cut. Anything that doesn't earn its slot is noise.

Cautionary lesson: difficulty cliffs are okay if they're teaching moments

Act 1 → Act 2 in Spire is a real difficulty cliff. The cliff is the test of "did you learn to take Elites?" Don't smooth out cliffs that exist for pedagogical reasons. Just make sure the cliff is teaching, not punishing for the sake of difficulty.