Every change to how VIRGIL grades.

A grade is only trustworthy if the method behind it is stable and, when it does change, the change is visible. VIRGIL logs every revision to its rubric and its grading calculation here — what changed, why, and how many grades moved. Nothing is altered silently.

Two kinds of entry appear below. A version change (v1.0, v1.1, …) is a change to the rubric itself — its signals, weights, or thresholds. A correction fixes how an existing signal is computed without changing the weights or thresholds. Both move grades, so both are recorded.

2026-06-20 · Creator history is read strictly as-of-launch

Correction (within v1.1 — no change to weights, bands, or thresholds). The all-time index of each wallet's prior launches had gone stale and was under-counting recent repeat launchers, so some serial deployers were being read as first-time founders. Two things were fixed: the index was made complete and self-maintaining, and the prior-launch count is now read as of each launch — only a wallet's launches that came beforea given launch count toward it. A deployer's later launches never reach back and change a grade already issued, and the count no longer shifts over time, so it reproduces exactly. Impact: 136 grades corrected — overwhelmingly mass-deployer spam that had been mislabeled, now reading its accurate F.

2026-06-13 · Rubric v1.1 — full deployer history

Version change.The serial-deployer signal moved from “launches VIRGIL had observed since it began” to the complete all-time index of every token a wallet has launched on the launchpad — the same scoring bands and points, on better data. (The 2026-06-20 correction above refines how this index is read.)

2026-06-12 · Rubric v1.0 — the initial instrument

Version change. The first public rubric: three weighted on-chain signals — launch configuration (40%), creator history (35%), and presence (25%) — combined into a 0–100 composite and a fixed letter grade. Deterministic and reproducible, with no model, sentiment, or discretion.

For the full scoring method behind these versions, see the methodology.