Our Own Audit Notes

01What we reviewed, and how

• The launch surface is deliberately tiny. One ERC-20 (deployed by SuperRare's audited factory via the Rare CLI — not our code) + one renderer+721 lens contract with a tiny mint footprint (33 hero 1/1s). The less we deploy, the less there is to get wrong.
• Reproducible economics. Every tokenomics number is derived by scripts/token-model.mjs — run it and re-derive the price schedule, slippage, allotments, and lifetime burn yourself.
• Adversarial modeling pass. The tokenomics were pressure-tested by an adversarial pass — multiple independent models, each checked by a skeptic — to catch hand-wavy claims and hidden assumptions.
• SuperRare Sepolia audit. A testnet rehearsal drew a five-item review from SuperRare; our point-by-point reply is in docs/AUDIT-REPLY.md. Highlights below.

02Things we found and fixed (in the open)

• The burn schedule overflowed the cap. An early card-retirement schedule totalled 4.36M burn against a 3.03M mint-once cap — impossible. Fixed by removing the mechanic: v2.2 cut forced card retirement / ash entirely. The burn is now pure token deflation (⅔ of the cap, ~2.02M), and the deck survives. The only rule left is the trivial Σ burns ≤ cap.
• Mint/burn assumption, corrected. We'd assumed burns re-mint on the next buy. SuperRare confirmed the opposite — minted once, burns permanent. The whole model was rebuilt on that (3× contraction).
• The lens standard. We'd sketched a custom ERC-1155. The cohort docs are clear: lenses are ERC-721. Fixed: every card is a render-by-id lens on one renderer+721 contract — 33 minted, 67 render-only, Lovebeing a holder lens — a tiny mint footprint, no 1155.
• Renderer display bugs. The render read 100% "circulating" and 0 per-RARE; fixed to read BURNED % (max − total) and a 2-dp per-RARE value in contracts/UR3030RenderPrototype.sol (re-callable setRenderContract, no token redeploy).
• A ~100× pricing error in an early doc. A stale peg implied a ~$0.001 pack. Fixed: the peg is ~1 RARE/token; the pack is a ~350-token bundle ≈ $7.
• Over-claimed "supply only goes down." Fixed: the docs now state plainly that net supply = buys − burns (sign indeterminate short-term); deflation is pressure, and over the deck's life burns dominate.

03Invariants we designed for (launch)

04What is still pending before mainnet

• Independent third-party security audit of the renderer+721 lens contract — not yet performed. (The ERC-20 + curve come from SuperRare's protocol; the render/lens is ours.)
• Curve calibration — open at ~1 RARE/token on the 3.03M supply (with SuperRare).
• Lens setup with SuperRare — does the assisted 721 setup support render-by-id across 100 card-lenses (33 minted, 67 render-only), or do we deploy our own combined renderer+721 lens contract via the CLI?
• Mint mechanism — a claim/voucher redeemer for the 11 gacha lenses (pack burn) and the 22 earned lenses (signed game vouchers).
• Full Sepolia dress rehearsal — deploy, wire the render, run a mock season: rips (buy+burn), a hero-lens mint, and watch the burn meter climb in the render.

05Verify us yourself

• Run the model — node scripts/token-model.mjs reproduces every number on the Tokenomics page.
• Watch the chain — packs and conviction are burn txs on the one token; hero-lens mints are wallet-signed; getMarketState() is live.
• Read the design — docs/ECONOMIC-FLOW.md (canonical) + docs/AUDIT-REPLY.md (the SuperRare reply).
• Read the render fix — contracts/UR3030RenderPrototype.sol (BURNED %, per-RARE).