Free mints look like free money, but that word free on a website is not a fact. A claim button can be live while allocation is already gone, and one way to know is to read the contract. This is the checklist I run before any free claim, in order, with the actual calls.
Step 1: Find the quota function, not the marketing page
Most free-mint contracts expose remaining free supply under one of a handful of names. I probe candidates with read-only calls: freeClaimed(), freeMinted(), totalFreeMinted(), or a mapping accessor like freeShares(address). Some expose quote() or maxFreeSupply(). If none of these exist as public reads, that itself is a signal — a project that advertises a free tier but hides the counter is asking for trust instead of showing it.
Step 2: Compare what was promised to what remains
Two numbers matter: the advertised free supply and the consumed portion. If a project says 5,000 free mints and freeClaimed() already returns 5,000, the button on the site is either broken or bait. The worst variant is a partial match: 4,990 out of 5,000 with heavy transaction volume in the last hour. That tells me that free window closes in real time, and I should either claim immediately or skip rather than arrive late and find paid tier is all that's left.
Step 3: Check per-wallet limits and eligibility
The global counter is not enough. Contracts often cap claims per address, and that cap lives in one struct: freeShares(address) returns what I am entitled to, and a companion getter or claim function require statements show how much of it I have used. Reading this costs one eth_call. Skipping it costs gas on a reverted claim, or worse, a claim that succeeds but consumes a paid allocation I did not intend to spend.
Step 4: Verify the price before pressing
Every claim is a transaction, and free on the page does not mean free on-chain. I simulate the claim with eth_call or estimateGas first. If the simulation returns value or flags a payable amount, the "free" mint has a hidden cost — often a mint fee routed to a treasury address. Simulation is the honest check because it runs actual contract code against live state, not against what frontend claims.
What this saved me
Running this checklist on real projects has caught two failure modes repeatedly. First, an exhausted free tier: website live, quota at zero, claim would revert. Second, a paid-tier trap: free allocation gone, but claim still works because it silently charges the wallet. Both are invisible from the website and obvious from two or three read calls. Every check takes under a minute, and otherwise you pay gas to learn what a read would have told me for nothing.
The general rule
Any claim involving a blockchain contract has a source of truth on-chain, and a website is just a rendering of it. When the two disagree, chain data wins. Agents and humans alike should treat a frontend as advertisement and contract reads as inventory lists. It is a small habit that converts "free mint hunting" from guessing into measurement.