The method, in full.

ProofGrade grades applications on Robinhood Chain from A to F. A grade answers one narrow question: does the thing an app claims actually exist in its contracts? That question has a right answer, it can be re-run by a stranger in a minute, and almost nobody on this chain is answering it.

We do not predict whether a project will succeed, assess a team's intent, or offer an opinion on price. Those are unfalsifiable, and unfalsifiable claims can be bought.

Each gate is a binary check against public chain state. No gate accepts a judgement call. Four are fatal: failing any one of them produces an F on its own, regardless of everything else.

any FATAL failed  →  F, immediately

otherwise, score = 100
  failed:        − 18 MAJOR    − 7 MINOR
  unreadable:    − 15 FATAL    − 8 MAJOR    − 4 MINOR

an unreadable FATAL gate caps the grade at C

A  90–100    B  75–89    C  60–74    D  40–59    F  < 40

Indeterminate is never silently treated as a pass. If a gate cannot be read, we say so and the score carries the penalty. Treating unknowns as passes is precisely how an audit process quietly becomes useless.

Every failed gate publishes the exact call that failed it: the query, the block height, and the value returned. This is the entire moat. A grade with the query attached is checkable, so it cannot be dismissed as an opinion — and it is a lie we could be caught in within a minute.

GATE 3 — CUSTODY PATH                      FAIL

Deposit destinations resolve to externally-owned
accounts, not contracts.

evidence:
  eth_getCode(0xAbC…1234) @ block 34,967,112  →  0x
  (empty bytecode = ordinary wallet, not a contract)

Reads are cross-checked against more than one RPC provider. A single lying endpoint must not be able to move a grade.

• We never accept payment from a project we grade, in any form, including for expedited review.
• We never write the word “scam”. We publish which gate failed and the evidence.
• We never grade intent, team quality, or whether a token will appreciate.
• We never take custody of user funds. The system is read-only by construction.
• We never quietly edit a grade. Corrections publish as new versions with the reason, and history stays public.

Our revenue comes only from people who need grades to be accurate: integrators protecting their own users, and holders protecting their own capital. Both are damaged the instant we inflate a grade.

Any graded project may dispute a finding. Appeals are published in full alongside our response — including when the appeal is upheld and we were wrong. Reports are append-only: a new grade supersedes an old one by reference, never overwriting it.

Being publicly corrected once, early, and handling it cleanly is worth more than an unblemished first month.

The token launches after the grader has a track record, not before. Our launch narrative is a teardown of projects that shipped a token first and a product later; doing the same thing ourselves would hand every critic the sentence we used on everyone else.