collinzrj

collinzrj / sts_lean

Public

Core logic of Slay the Spire implemented in Lean4 for formal verification proofs

19
1
100% credibility
Found Apr 11, 2026 at 19 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Lean
AI Summary

A benchmark project that uses formal math proofs to verify infinite damage combos in the card game Slay the Spire, designed to test AI assistants on solving complex game logic puzzles.

How It Works

1
🎮 Discover the magic

While playing Slay the Spire, you hear about a fun project that mathematically proves if your favorite card combos can loop forever and deal endless damage.

2
📥 Grab the toolkit

Download the project files to your computer and run a quick setup to prepare ready-to-solve puzzle challenges about game combos.

3
🤖 Challenge an AI

Hand the puzzles to a smart AI assistant, like Claude, and ask it to fill in the proofs showing the combos work perfectly.

4
✨ See proofs come alive

Watch excitedly as the AI crafts detailed step-by-step proofs that confirm the combos loop back to the start while dealing damage every time.

5
🔍 Check the work

Run the built-in grader to verify everything matches up, ensuring no cheating and all proofs are solid without shortcuts.

🏆 Proofs confirmed!

Celebrate as the results show which combos are truly infinite, even against tricky card shuffles, unlocking deeper game understanding.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 19 to 19 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is sts_lean?

sts_lean implements Slay the Spire's core logic—drawing, shuffling, discarding, and 90+ card effects—in Lean 4 for formal proofs of infinite combos. It lets you verify loops that deal unbounded damage, from lucky draws (existence proofs) to adversarial shuffles controlled by oracles. Run lake build to check proofs, or use Python scripts to prepare benchmarks and grade LLM attempts on core logic verification.

Why is it gaining traction?

It bridges game mechanics with theorem proving, offering tiered challenges: simple traces for existence, full tactics against shuffle oracles. Eval scripts snapshot files, build modules, and audit axioms, delivering tamper-proof LLM scores. Niche but sharp for testing core logics in Lean, outpacing ad-hoc proofs or simulations.

Who should use this?

Lean users formalizing game engines or infinite loops. AI researchers benchmarking LLMs on proof synthesis via GitHub core actions. Slay the Spire combo hunters seeking rigorous validation of leangains-style decks.

Verdict

Worth forking for Lean experiments or LLM evals—benchmark shines despite 19 stars and 1.0% credibility score. Docs cover commands and limitations; maturity lags but core setup delivers immediate value.

(178 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.