What does a stablecoin dashboard prove when its numbers do not move? Not as much as a green status badge might imply.
I built Mezo MUSD Proof Desk, an independent, open-source, AI-assisted read-only checker for selected MUSD and MEZO contracts and documented pools on Mezo mainnet. It uses Python's standard library to collect block-pinned public chain evidence and publishes static JSON that anyone can inspect. It does not ask for a wallet connection.
Here is a small reproducible comparison—and the important limits of its conclusions.
The checker pinned contract reads to two different mainnet blocks on chain ID 31612:
| Observation | Block | Timestamp (UTC) |
|---|---|---|
| First | `12709282` | 2026-10-23 14:49:48 |
| Second | `12709416` | 2026-10-23 14:58:23 |
The first block hash was 0xec725d...08528d9, and the second was 0x5c398b...64a81d. The full hashes, contract addresses, read method, and comparison record are in the public source and pinned evidence.
For both samples, these on-chain values were unchanged:
| Item | Observed value at each block |
|---|---|
| MUSD ERC-20 totalSupply | 29,482,016.986182972908040124 |
| MEZO ERC-20 totalSupply | 1,138,540,989.748240837394828695 |
| MUSD / BTC pool's MUSD reserve | 39,338.597921228749227732 MUSD |
| MUSD / mUSDC pool's MUSD reserve | 6,852.662834954862377028 MUSD |
| MUSD / mUSDT pool's MUSD reserve | 7,700.365872321640162678 MUSD |
The three selected pools held a combined 53,891.626628505251767438 MUSD in the sampled reserve fields. This figure is not total protocol TVL, dollars of executable liquidity, redeemable collateral, or a measure of MUSD backing.
Runtime-bytecode SHA-256 fingerprints for the sampled contracts and pools were also unchanged between these blocks. A byte-for-byte hash match is not a source-code audit or a security guarantee.
The proof collector follows a narrow process:
eth_chainId before any state read. A different chain is an error.token0 / token1; read getReserves and MUSD balanceOf at that same block.decimals and totalSupply without silently assuming every token uses the same decimals.
The original repository contains 40 offline regression tests across the collector and sample-report generator. These exercise validation and failure handling; passing them does not establish that the underlying contracts are safe.
"The stablecoin is fully backed." The shown ERC-20 supply and pool reserves say nothing by themselves about BTC collateral, outstanding liabilities, redemption capacity, or the peg.
"A swap of this size would execute." Stable-pool pricing depends on its invariant, fees, route, amount, changing balances, and other protocol details. A reserve is not a trade quote.
"The protocol was inactive." Two blocks with identical sampled values do not rule out events in other blocks, different contracts, bridged representations, or off-chain changes.
"This is a security audit or yield recommendation." No exploits, solvency model, price feed, borrow path, or vault APY were analyzed. The tool sends no transactions, pays no gas, and holds no customer assets.
That separation—evidence observed versus economic inference unproven—is the point of the exercise.
Interested in a scoped paid QA comparison? I can propose a human-reviewed, source-only report that compares two named public Mezo blocks for a documented MUSD contract/pool set. The 19 native Base Circle USDC figure on the demo is my seller quote, not an accepted order or a claimed past sale. Scope, provenance, AI-assisted review, IP, exact chain/token/payee, legal suitability, and payment/acceptance terms must be agreed before any custom work. Reach me at onecursor1@gmail.com.
Disclosure: independently operated, AI-assisted original analysis by one human owner; not affiliated with Mezo. No investor recommendation, peg/backing attestation, customer trade, commissioned audit, contest award, or verified income is implied.