Blockchain

What onchain data can and cannot prove

A chain is a receipt book, not a witness statement. It proves what moved and which code ran; it does not prove who meant it or what they promised off-chain. Keeping those apart is the difference between analysis and astrology.

Short answer: Onchain data proves movement and code execution; identity, intent and off-chain promises are inferences that need a source and a confidence.

What the chain proves

Balances at a block, transfers, approvals, contract calls and the code that executed. If a transaction is final, the movement happened — that part is not an opinion.

What it only suggests

Identity, intent and off-chain agreements are inferences. A label on an address is a probabilistic claim with a source and a confidence, and a clean address is not proof of an honest counterparty.

Holder concentration, wallet clustering and "smart money" tags are heuristics. They are useful when the confidence is shown and dangerous when it is hidden.

Four checks that catch expensive mistakes

Before signing anything, these four cover most of the loss:

How GTXAgent covers it

Each check is a paid call with a fixed price: a token verdict, a transaction preflight with decoded balance changes, and a counterparty dossier with every label carrying its source and confidence. The assistant combines them into one decision when a trade is being considered.

Run the check on the actual contract or address, not on a screenshot.

Related

Questions

Can onchain analysis identify a thief?

Sometimes the flow is traceable to an exchange deposit or a known cluster, but attribution is probabilistic. The chain proves the movement; the identity needs an off-chain source.

Is holder concentration reliable?

It is a heuristic and is unavailable on some tokens. GTXAgent names it as a gap instead of estimating it.