How the instrument works.

01What PROBUS is

An instrument for reading Solana mints. You give it an address; it reads the mint account, the two authorities, the supply, and every pair an indexer knows about, then prints each figure beside the thing that figure does not cover. It does not score, grade, rank or recommend. Read the chain. Print the limits.

02The three-state rule

Every field is a value from a named source, a measured zero, or n/a with the reason attached. Those three never share a cell.

This is the whole point. Most tools in this category return $0 when a liquidity call fails, and on the page that reads exactly like a pool that was drained — the reader cannot tell the difference, and nothing warns them. PROBUS renders n/a in grey italic with the reason on hover, so a gap in the instrument can never be mistaken for a finding about the token.

03The authorities

The mint authority and the freeze authority are the two most consequential things here. A live mint authority means supply can still be increased; a live freeze authority means balances can be frozen. Both are reported as addresses, not as a tick or a cross — who holds one is a thing you may want to go and look at, and a boolean would deny you that. PROBUS does not assess the holder.

04The anchor

An instrument can be wrong in a way nobody notices: a stale account, a mis-read field, a provider quietly substituting. Every figure would still look plausible. So PROBUS re-reads two mints whose properties it does not own — USDC, whose mint and freeze authorities are both live at published addresses, and wSOL, where both are revoked. Opposite directions, so a reader that invents an authority and one that loses one are each caught by a different subject.

It runs at boot and hourly, and the result rides the header of every sheet. It turns red the moment a field stops agreeing. It covers the mint reader and nothing else — market figures come from a third party and have no fixed answer to check against, which is stated on the anchor sheet rather than glossed over.

05Where the numbers come from

Chain state from a Solana RPC, held as a list with failover — a single endpoint going quiet has taken a project down before, so whichever answered is remembered and named on the page. Market state from DexScreener: pairs, liquidity, volume, and when the oldest pair was created.

06Concentration, and why it may say n/a

Holder concentration needs getTokenLargestAccounts, and every free RPC refuses it — the public endpoints answer "too many requests for a specific RPC call" or ask for a paid plan. Without an indexer key the column reads n/a with that reason on every row, rather than disappearing. This deployment has one configured.

Even with a key it fails on the very largest mints: USDC and wSOL come back too many accounts requested, because ten million holders is more than the call will serve. That is a permanent limit, and the n/a says so rather than pretending the reading is merely pending.

When it does work, accounts owned by known AMM and bonding-curve programs are resolved and excluded from the share. A pool vault holding 40% of supply is not a whale, and counting it as one is the commonest mistake in this category — one small mint tested during the build had 93% of supply sitting in its bonding curve, which unexcluded would have read as a single holder owning almost everything.

07Freshness

The archive answers from store unless liquidity has drifted more than a tenth, or the record is over a day old. A re-scan that returns the same numbers spends an upstream call and teaches nobody anything. Every answer says whether it came from the archive and how old it is, and you can always force a fresh read.

08The archive

Every scan ever run is kept, and nothing is edited afterwards. A re-scan appends a new reading rather than replacing the old one — a record that could be rewritten later would be worthless as evidence, and it means a figure can be watched moving instead of being handed to you as today's number alone.

09What this is not

Not advice, not a rating, not a signal. There is no score anywhere on this site and no colour that means "good" — only one that means this can still change and one that means this could not be read. A clean reading is not a recommendation; it is a description of four or five public facts at one moment.