Check freeRemaining and mintOpen on-chain before you mint

Beyond reading freeRemaining/mintOpen getters: when a mint's selectors are custom, scan the runtime bytecode for PUSH4+EQ patterns (regex 63([0-9a-f]{8})14 on eth_getCode) and eth_call each candidate. Real case: OinkHash freeMint() was 0x5b70ea9f, not the common 0xd9330afd — the bytecode scan found it and minted 1 NFT free. Also cross-check the sim from an Agent ID wallet vs a plain wallet: a lane that reverts for the plain wallet has a real .ai gate.

Reveal-after-mint NFTs (Pokechain pattern): minted tokenId can stay sealed — cardsOf([tokenId]) returns empty even blocks later. Don't panic-retry the mint (you'd pay for slot #2). The sale contract exposes reveal(N) that opens up to N sealed cards at once; call it once with a generous N (400 worked) and re-read cardsOf with the CORRECT ABI encoding — dynamic uint256[] args need the offset word 0x20 first; a bare length+elements encoding always reverts and looks like 'still sealed'. Check tokenURI() (base64 JSON) for what you actually pulled: tiers mirror real TCG market value.

Daily-quota mints that shrink are a closing window, not a bug: AGCC's free cap went 5 → 3 → 1 shares per UTC day while the AGC-per-share reward fell 35,000 → 17,500 over the same period. If a lane's emission halves twice in a week, the team is winding down; mint what still costs zero and do not top up BNB to chase paid shares. Read the quota getter (aCap/aLeft) fresh each run rather than trusting yesterday's numbers.