# Neon Durable Current-Target Contract Catch

Date: 2026-05-24

Target: LilArcade Neon Step Sequencer, Monkberry Moon Delight (Tab), mix profile `monkberry-moon-delight-eq-lane-mix-v7`.

## Claim

A durable approved mix override should be provable on the current production target through the live proof contract, the source/profile descriptor, visible UI text, and basic offline render guardrails.

## Finding

The first production run of the new durable current-target proof failed even though the app had the right level in the places a normal smoke proof would check:

- selected song matched `Monkberry Moon Delight (Tab)`
- mix profile id matched `monkberry-moon-delight-eq-lane-mix-v7`
- live contract level for `chord` was `0.18`
- visible UI showed `0.18X`

The failing receipt showed `profileLevel: null`, so the proof could not verify that the durable source/profile descriptor agreed with the live mixer state.

This was a proof-contract gap, not a musical-taste finding. The mix was not declared better or worse. The proof only said the durable-current-target claim was not fully supported yet.

## Fix

LilArcade PR #501 added `npm run proof:sequencer:durable-current-target`, exposed `mixProfile.mixerLevels` through Neon proof state, mixer state, and offline metric receipts, and taught the summary parser to read stored `window_eval` returns by `return_stored_to`.

## Final Receipt

The final production proof passed with zero findings:

- expected `chord: 0.18`
- actual live level `0.18`
- profile descriptor level `0.18`
- visible token `0.18X`
- peak `0.7777`
- RMS `0.1112`
- clipping `false`
- low-level window `false`

Boundary: this proves deterministic current-target and proof-contract receipts for the durable override. It does not prove subjective mix taste.