# Evidence for [#11873](https://git.cafe/pingdotgg/t3code/issues/11873)

Captured all six matrix rows: **S1, S2, S3, R1, R2, R3**. Twelve static grid pairs plus one narrow chat pair (**26 original stills**), **26 diff images**, **7 comparison sheets**, and **6 original recordings plus 1 supplementary recording**. No required row omitted. These replace the preliminary uploads, which are not coverage proof.

BEFORE: [5ffe045931](https://git.cafe/pingdotgg/t3code/commit/5ffe04593166870d3ada2a83eb5fe55f0d192fc7) (`5ffe04593166870d3ada2a83eb5fe55f0d192fc7`). AFTER: [330326c7db](https://git.cafe/pingdotgg/t3code/commit/330326c7db0bbc8165d550c9c5082d13b7dee9d7) (`330326c7db0bbc8165d550c9c5082d13b7dee9d7`). Main was merged without conflicts in [c9f4d75501](https://git.cafe/pingdotgg/t3code/commit/c9f4d755017785285e2e923f74d260959823eab7). The baseline moved from the original `3b95e7831b` to current main at maintainer direction, before any accepted captures. All AFTER evidence uses the same final SHA with committed fixture/seeder files. At original capture completion the worktree was clean on `fix/issue-11873`; the supplement and final-head timing are recorded below.

## Coverage and sheets

Each sheet retains complete before/after frames and both diff maps. Desktop sheets contain light/dark rows at **1440×1000 CSS px**; narrow sheets contain light/dark rows at **390×844 CSS px**. R1 endpoint is 390×844 dark. Columns are labeled BEFORE, AFTER, zero-threshold absolute RGB difference **amplified 8× with saturation**, and **max RGB delta >8/255** changed-pixel map. No cropping or privacy masks were applied to originals.

| Row | BEFORE SHA | AFTER SHA | Outcome | Sheets |
| --- | --- | --- | --- | --- |
| S1 | [5ffe045931](https://git.cafe/pingdotgg/t3code/commit/5ffe04593166870d3ada2a83eb5fe55f0d192fc7) | [330326c7db](https://git.cafe/pingdotgg/t3code/commit/330326c7db0bbc8165d550c9c5082d13b7dee9d7) | Authentic Codex 0.163.0, empty isolated home: probe-failure notice becomes existing empty state. Ready status/models retained. | [1440](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/5d39a6c8cbcee6cb/S1-1440-sheet.png) · [390](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/6c6151a02b846804/S1-390-sheet.png) |
| S2 | [5ffe045931](https://git.cafe/pingdotgg/t3code/commit/5ffe04593166870d3ada2a83eb5fe55f0d192fc7) | [330326c7db](https://git.cafe/pingdotgg/t3code/commit/330326c7db0bbc8165d550c9c5082d13b7dee9d7) | Same authentic Codex plus independent CLI Proxy protocol source fixture: only failure notice disappears; source windows remain. | [1440](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/f9444e7f585789f7/S2-1440-sheet.png) · [390](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/9ce5f90f28d4a4df/S2-390-sheet.png) |
| S3 | [5ffe045931](https://git.cafe/pingdotgg/t3code/commit/5ffe04593166870d3ada2a83eb5fe55f0d192fc7) | [330326c7db](https://git.cafe/pingdotgg/t3code/commit/330326c7db0bbc8165d550c9c5082d13b7dee9d7) | Accountless/false-auth native-success protocol fixture in real T3: identical windows in all four cells; no real subscription login. | [1440](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/7f3c930f7e969ecc/S3-1440-sheet.png) · [390](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/66cefed27bc6b8fa/S3-390-sheet.png) |
| R1 | [5ffe045931](https://git.cafe/pingdotgg/t3code/commit/5ffe04593166870d3ada2a83eb5fe55f0d192fc7) | [330326c7db](https://git.cafe/pingdotgg/t3code/commit/330326c7db0bbc8165d550c9c5082d13b7dee9d7) | Codex selectable; typed /usage chooses /usage-limits; report opens, retains two proxy accounts, changes native account text, dismisses; selector reopens in the supplementary AFTER clip. | [390](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/7d7ccad4599cc47c/R1-390-sheet.png) |
| R2 | [5ffe045931](https://git.cafe/pingdotgg/t3code/commit/5ffe04593166870d3ada2a83eb5fe55f0d192fc7) | [330326c7db](https://git.cafe/pingdotgg/t3code/commit/330326c7db0bbc8165d550c9c5082d13b7dee9d7) | Accountless/false-auth refresh recovery: stable native-demo accountId; 12% used → unrelated -32000 failure retains windows/usage checkedAt → 23% used with newer usage checkedAt. No reload. | Recordings and published snapshots below |
| R3 | [5ffe045931](https://git.cafe/pingdotgg/t3code/commit/5ffe04593166870d3ada2a83eb5fe55f0d192fc7) | [330326c7db](https://git.cafe/pingdotgg/t3code/commit/330326c7db0bbc8165d550c9c5082d13b7dee9d7) | Accountless/false-auth exact refusal → success on later explicit refresh. ProbeFailed BEFORE / unsupported AFTER; both recover. No reload or negative cache. | Recordings and published snapshots below |

S1/S2 use the **real installed Codex 0.163.0 executable**, configured with a fresh empty CODEX_HOME and a dummy proxy token. The usage refusal occurs before an upstream request; these captures do not prove proxy turn connectivity. S2/R1's independent usage source implements the repository adapter's `/v0/management/auth-files` and `/api-call` envelope with two fake accounts (12%/38% used), fixed reset timestamps and no real secrets. S3/R2/R3 use a clearly identified **newline-delimited app-server JSON-RPC protocol-boundary fixture in the real T3 app**, not a real native subscription login and not injected client state. The peer echoes request ids, returns null account and false requiresOpenaiAuth, selectable model discovery, and explicit success/auth/transient phases. Phase changes occur only through the external control file.

## Interaction recordings — 1440×1000 CSS, light

R2 is **refresh recovery**, not streamed notification or live-turn verification. Published snapshots were fetched from the real server's `server.getConfig` RPC over an authenticated browser WebSocket after the observed refresh milestones. Auth/server clocks remained real. The protocol trace and published snapshots corroborate the recording; a still cannot prove this transition.

R1 script: open the Codex provider/model selector, select Codex, type `/usage`, choose `/usage-limits`, inspect the report and its two independent proxy accounts, dismiss the report, then reopen the selector. Observed original AFTER milestones from proof audit: selector visible **6.000s**; report visible **35.500s**; report dismissed by **36.240s**; recording ends **36.276333s**, before selector reopen. Original BEFORE shows the reopened selector at **4.800s** (duration **4.985900s**). The original AFTER alone does not prove the last step.

[AFTER supplementary dismissal → reopen](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/692bac30d3df4ec4/R1-after-dismiss-reopen-1440-light.mp4) supplies only that missing ending, with its own zero-based clock: report visible at **0.000s** and **6.400s**; dismissed by **6.600s**; selector visibly open by **7.400s**, remaining open at **18.000s**; total duration **19.104233s**. Those are inspected video-frame times, not inferred click timestamps. Report retains `demo-a.json` and `demo-b.json`; reopened menu visibly lists Codex (via cliproxy) models. The menu was verified in both the live snapshot and actual video before acceptance. This is a separate continuous recording, not a splice into the original.

Supplement provenance: committed `seed.py R1`, real installed Codex and committed source fixture, same isolated worktree `.t3`, same Chrome 154/Linux, **1440×1000 CSS, DPR2, light**, browser Date fixed **2026-10-22T12:00:00Z**, fonts loaded, sidebar/details closed. MP4 **2880×2000**. Recorded **08:58:48.059–08:59:07.012 UTC** at running head [330326c7db](https://git.cafe/pingdotgg/t3code/commit/330326c7db0bbc8165d550c9c5082d13b7dee9d7). After capture/teardown, the CI agent committed fixture import/reference formatting as [92e1f121c9](https://git.cafe/pingdotgg/t3code/commit/92e1f121c9bc12fead91a51e5a8406d83960130e) at **08:59:34 UTC**; final checkout is clean at that newer head. This supplement does not claim a capture at the newer head. No production edits, commits, pushes or forge writes were performed by the evidence agent.

| Row | BEFORE SHA | BEFORE recording | AFTER SHA | AFTER recording |
| --- | --- | --- | --- | --- |
| R1 | [5ffe045931](https://git.cafe/pingdotgg/t3code/commit/5ffe04593166870d3ada2a83eb5fe55f0d192fc7) | [BEFORE](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/bf9042a40437cb18/R1-before-1440-light.mp4) | [330326c7db](https://git.cafe/pingdotgg/t3code/commit/330326c7db0bbc8165d550c9c5082d13b7dee9d7) | [AFTER](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/e2f783283cfc83f8/R1-after-1440-light.mp4) |
| R2 | [5ffe045931](https://git.cafe/pingdotgg/t3code/commit/5ffe04593166870d3ada2a83eb5fe55f0d192fc7) | [BEFORE](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/3c15aff0a86f1445/R2-before-1440-light.mp4) | [330326c7db](https://git.cafe/pingdotgg/t3code/commit/330326c7db0bbc8165d550c9c5082d13b7dee9d7) | [AFTER](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/3e7a59b8e34eaaa0/R2-after-1440-light.mp4) |
| R3 | [5ffe045931](https://git.cafe/pingdotgg/t3code/commit/5ffe04593166870d3ada2a83eb5fe55f0d192fc7) | [BEFORE](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/d9a4c71b3b3bc528/R3-before-1440-light.mp4) | [330326c7db](https://git.cafe/pingdotgg/t3code/commit/330326c7db0bbc8165d550c9c5082d13b7dee9d7) | [AFTER](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/b86707e39c913b8f/R3-after-1440-light.mp4) |

[Published R2/R3 snapshots](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/2f301ff46923bce2/published-snapshots.json) · [Authentic S1 protocol trace](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/5cd33f4a30ce8a13/S1-authentic-protocol.json)

## Rendering environment and timing

T3 collaborative server browser, Google Chrome / Chromium **154.0.8037.92**, Linux x86_64 (UA platform Linux; high-entropy platformVersion 6.1.102). **DPR 2**, browser zoom 1. Freeform viewport resizing preserves desktop user-agent and has no native-phone/touch emulation. A 390px web capture is responsive web evidence, not native mobile verification. DPR 2 was explicitly accepted by the maintainer in place of the auditor's DPR 1 assumption.

Whole-viewport PNGs: **1280×889** for the 1440×1000 CSS desktop layout (preview caps width at 1280; scale is 1280/2880 of the 2× raster, with rounded height). Narrow PNGs are **780×1688**, the unscaled 2× raster of 390×844 CSS. Both sides use identical scaling. Recordings are **2880×2000** MP4. The CSS body stack is `-apple-system, BlinkMacSystemFont, "Segoe UI", system-ui, sans-serif`; host fontconfig resolves generic system/sans and mono to Liberation Mono. Liberation Sans, Sans Narrow and Serif are also installed. No web fonts were pending (`document.fonts.status=loaded`).

Page JavaScript Date/new Date()/Date.now() fixed to **2026-10-22T12:00:00.000Z**. Fixed fixture reset times are 2026-10-22T16:00:00Z and 2026-10-29T12:00:00Z. Browser-internal HTTP/cookie auth clocks and server clocks were untouched. Refreshes set app-derived limits/report times after the freeze. Same benign empty thread/project context, selected provider, closed sidebar/details panel, top scroll, theme, viewport, fonts and rendering scale were used for comparisons. Finite DOM animations/font readiness and actual RPC/UI milestones were awaited; no sleeps were used to assert behavior. Originals contain no private conversation. The empty home has no auth.json.

## Pixel audit

Opaque PNGs are converted identically to RGB; alpha is discarded identically. Each channel's absolute delta is multiplied by 8 and saturated to 255 for the zero-threshold amplified image. The second map marks a pixel white only when max absolute RGB delta **strictly exceeds 8/255**. Raw originals are preserved; the amplified map is explicitly not raw1. Percentages below use all saved-frame pixels, without excluding intended changes.

| Pair | PNG size | >0 pixels (%) | >8 pixels (%) | Outside declared notice/reflow >8 |
| --- | --- | --- | --- | --- |
| R1-390-dark | 780×1688 | 329183 (25.001747%) | 32616 (2.477215%) | 0 |
| S1-1440-dark | 1280×889 | 36331 (3.192755%) | 36322 (3.191964%) | 0 |
| S1-1440-light | 1280×889 | 36363 (3.195567%) | 36331 (3.192755%) | 9 |
| S1-390-dark | 780×1688 | 91816 (6.973508%) | 91800 (6.972293%) | 0 |
| S1-390-light | 780×1688 | 91816 (6.973508%) | 91800 (6.972293%) | 0 |
| S2-1440-dark | 1280×889 | 36327 (3.192404%) | 36316 (3.191437%) | 0 |
| S2-1440-light | 1280×889 | 36327 (3.192404%) | 36316 (3.191437%) | 0 |
| S2-390-dark | 780×1688 | 91792 (6.971686%) | 91774 (6.970318%) | 0 |
| S2-390-light | 780×1688 | 91792 (6.971686%) | 91774 (6.970318%) | 0 |
| S3-1440-dark | 1280×889 | 0 (0.0%) | 0 (0.0%) | 0 |
| S3-1440-light | 1280×889 | 0 (0.0%) | 0 (0.0%) | 0 |
| S3-390-dark | 780×1688 | 0 (0.0%) | 0 (0.0%) | 0 |
| S3-390-light | 780×1688 | 0 (0.0%) | 0 (0.0%) | 0 |

**S3 is full-frame zero at both thresholds in all four cells.** S2's independent source controls have zero changed pixels above 8 outside the removed-notice region. R1's changes are confined to the declared account-text/reflow region. S1 desktop light retains **9 pixels above 8** at the sidebar glyph outside the notice; repeated same-build snapshots varied at this tiny glyph (3 then 9 pixels), so this is recorded as capture/raster variance, not an assertion of exact control equality. The threshold was not raised. Other S1 cells have zero outside the notice above threshold.

Declared comparison regions, in **saved PNG pixels**: S1 desktop y60–119 / narrow y140–299; S2 desktop y330–389 / narrow y1130–1289; R1 narrow y850–1389. These regions affect only statistics; no pixels are removed from either diff or original. There is no equality claim across intentional notice/empty-state or account-text reflow.

## Final diff, surfaces, checks

Re-read final production diff against the chosen main baseline: `CodexAppServerRequestError`, exact code **-32600**, exact message **codex account authentication required to read rate limits**, then account=null and requiresOpenaiAuth=false. No broad failure-string predicate. Success and unrelated failures keep their paths.

Shared render paths inspected: pooled web Usage uses shared notice collection that skips unsupported; chat keeps the account and renders the existing no-subscription notice. Desktop wraps these web paths. Mobile pooled usage and subscription widget share unsupported filtering; native mobile/Electron shell were not launched or visually verified. Typed command and command menu converge on the report; R1 records typing the abbreviation `/usage` and selecting `/usage-limits`. Provider/settings configuration remains available and model selection is recorded. No transport contracts, relay/tunnel/origin routing, other provider adapters, auth UI, scheduling or native managed-account flow changed. No new reverse-state action is introduced.

Focused validation on the final merged SHA: `vp test run src/provider/codexUsageLimits.test.ts src/provider/ProviderRegistry.test.ts` from apps/server: **2 files, 54 tests passed**. No repository-wide checks. No fixture-mirroring tests added. Both sides' R2 snapshot equality/identity/newer checkedAt and R3 classification/recovery were also checked directly against captured published snapshots. Existing provider-core tests (lead's scope) cover live-window acceptance after probeFailed; these recordings make no streamed-turn claim.

What else changed: the branch merged current main, then added only three reusable evidence files under `apps/server/src/provider/testFixtures/usageEvidence/`: `peer.mjs`, `data.json`, `seed.py`. Relative to current main, the existing four implementation/test files plus these three fixtures are the complete seven-file diff. No additional production changes, images, plans or manifests were committed. The script/data were committed **before** final capture; a byte-identical copy under /tmp allowed the same fixture to run on unmodified main. Seeder writes only worktree `.t3/userdata`, rejects the canonical live-userdata path and requires a worktree .git; direct writes were made only while the isolated server was stopped. Never started a server against live userdata. No subscription credentials accessed. No PR created, pushed or publicly commented.

## Fixture complexity and malformed-input timing

No custom parser or regex was introduced. Native `JSON.parse` is used for static data, the phase file, each newline-delimited JSON-RPC message and the local source request body. Parsing/serialization is O(n) time and O(n) space in the parsed/serialized input size; newline framing retains O(longest line) memory and handles each message once. Per-message dispatch, fixed account/model/window construction and phase selection are constant-size; emitted reply work is O(reply bytes). Control-file reads are constant-size for this fixture. Milestone append is O(serialized response bytes).

The source's `for await` body accumulator visits each chunk once, retains O(B) body bytes and has **O(B²) worst-case copying** for repeated immutable-string concatenation (V8 can use ropes); it is a disposable loopback fixture serving small deterministic management requests, not a production endpoint. The seeder enumerates T SQLite table names once (O(T), O(T) fetched metadata), checks fixed prefixes/one fixed scenario enum, deletes N projected rows (SQLite index maintenance can be O(N log N)), and inserts fixed-size safe context. JSON dump/write work is linear in the fixed payloads. No event history is used as behavior evidence.

Quick malformed-input benchmark: 3,000 JSON lexical value tokens (2,999 `0` values followed by malformed `!`, **6,001 bytes**), sent as one JSON-RPC line to the **actual peer executable**, five fresh Node processes. Each failed immediately with nonzero exit, no hang. End-to-end including Node startup: **81.515, 73.091, 78.810, 68.095, 73.185 ms** (median **73.185 ms**). This measures fixture rejection, not production parser throughput; malformed HTTP-body timing was not separately measured.

[Malformed-input timing data](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/80e68577f293b39b/parser-timing.json)

## Missing / unverified

No matrix row is missing. No real subscription login/upstream success, native mobile app, Electron-specific shell behavior, remote relay/tunnel matrix, or streamed live turn was verified in this capture pass. S1 light's 9-pixel sidebar variance prevents an exact outside-region equality claim for that cell. Narrow Usage header crowding and abbreviated source labels are existing shared-layout behavior visible in both versions; no production UI edits were made for evidence.

All dev/source servers started for evidence were stopped via retained sessions; no listeners remained on 6706, 14746 or 18317 at final check. Final evidence-repair checkout is clean on `fix/issue-11873` at the later CI fixture-formatting commit described above. The capture running head remains the stated AFTER SHA.

## FRICTION

- recording repair | original AFTER ended before selector reopen | preserved original, added a separately timed continuous dismissal-to-reopen supplement; verified visible open menu before stopping.
- concurrent checkout | CI agent advanced the branch after capture/teardown | recorded capture head and later final head separately; no reset or recapture of accepted matrix.
- startup | migration-history warning lists three newer peer-link migrations | isolated server recovered with zero pending work; R1 flow succeeded.

- lifecycle | inherited dev server was no longer listening | started tracked replacement sessions and stopped each before fixture writes.
- fixture schema | first disposable seed omitted required thread fields | corrected canonical creation/lineage/model selection before choosing the final capture SHA.
- browser | preview renders at DPR2 and caps desktop PNG width | retained identical accepted DPR/scale and reported measured CSS/raster dimensions.
- capture | initial endpoint screenshot preceded finite panel animation | awaited animation/font completion, recaptured and inspected the actual PNG.
- capture | tiny S1 light sidebar glyph varies across repeated raster captures | preserved originals, reported 9 above-threshold pixels, did not raise tolerance or mask pixels.
- source labels | pooled usage represents accounts with avatar segments, not full email text | R1 report proves independent source labels/accounts remain.
- credentials | no native subscription login is available or authorized | labeled S3/R2/R3 protocol fixtures honestly; genuine S1 uses empty home.
- recovery | probe responses do not prove streaming/live-turn behavior | labeled R2 refresh recovery and attached actual published snapshots.

Model/harness: gpt-6.1-sol through Codex in T3 Code.

## Downloadable index and environment

[upload-markdown.md](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/4824c77beaf8019c/upload-markdown.md)
[rendering-environment.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/96def6c612fd65b7/rendering-environment.json)

## Verbatim ghdrop asset index

![R1-390-dark-diff-amp8.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/c9c912a61f2d7fc9/R1-390-dark-diff-amp8.png)
![R1-390-dark-diff-threshold8.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/00b0f758f87e0574/R1-390-dark-diff-threshold8.png)
![R1-390-sheet.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/7d7ccad4599cc47c/R1-390-sheet.png)
[R1-after-1440-light.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/e2f783283cfc83f8/R1-after-1440-light.mp4)
![R1-after-390-dark.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/743c9d4fc59f124b/R1-after-390-dark.png)
[R1-before-1440-light.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/bf9042a40437cb18/R1-before-1440-light.mp4)
![R1-before-390-dark.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/4b312496ee0b23f1/R1-before-390-dark.png)
[R2-after-1440-light.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/3e7a59b8e34eaaa0/R2-after-1440-light.mp4)
[R2-before-1440-light.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/3c15aff0a86f1445/R2-before-1440-light.mp4)
[R3-after-1440-light.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/b86707e39c913b8f/R3-after-1440-light.mp4)
[R3-before-1440-light.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/d9a4c71b3b3bc528/R3-before-1440-light.mp4)
![S1-1440-dark-diff-amp8.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/bba08016a95d6c4e/S1-1440-dark-diff-amp8.png)
![S1-1440-dark-diff-threshold8.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/83d72f7dafde7fcb/S1-1440-dark-diff-threshold8.png)
![S1-1440-light-diff-amp8.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/33b11a923ebe04d1/S1-1440-light-diff-amp8.png)
![S1-1440-light-diff-threshold8.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/ea293c4b7958345d/S1-1440-light-diff-threshold8.png)
![S1-1440-sheet.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/5d39a6c8cbcee6cb/S1-1440-sheet.png)
![S1-390-dark-diff-amp8.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/b563b88ed7241f77/S1-390-dark-diff-amp8.png)
![S1-390-dark-diff-threshold8.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/93ed13d1c5dfe036/S1-390-dark-diff-threshold8.png)
![S1-390-light-diff-amp8.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/9844b2c0d8754bc5/S1-390-light-diff-amp8.png)
![S1-390-light-diff-threshold8.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/a4198e20e125c269/S1-390-light-diff-threshold8.png)
![S1-390-sheet.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/6c6151a02b846804/S1-390-sheet.png)
![S1-after-1440-dark.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/402b95f8bec154f6/S1-after-1440-dark.png)
![S1-after-1440-light.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/5bea1d4a13d65b4b/S1-after-1440-light.png)
![S1-after-390-dark.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/ddccf921cfc7320b/S1-after-390-dark.png)
![S1-after-390-light.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/923726b521a1c449/S1-after-390-light.png)
[S1-authentic-protocol.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/5cd33f4a30ce8a13/S1-authentic-protocol.json)
![S1-before-1440-dark.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/47833d271a9148f1/S1-before-1440-dark.png)
![S1-before-1440-light.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/e455f783b17a8ead/S1-before-1440-light.png)
![S1-before-390-dark.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/5c7f8c89d3585a95/S1-before-390-dark.png)
![S1-before-390-light.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/b3f311dbf39273c7/S1-before-390-light.png)
![S2-1440-dark-diff-amp8.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/3d7e6ab9387ccc6a/S2-1440-dark-diff-amp8.png)
![S2-1440-dark-diff-threshold8.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/ec1e6c1bcd2866d9/S2-1440-dark-diff-threshold8.png)
![S2-1440-light-diff-amp8.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/00eea166bbcac655/S2-1440-light-diff-amp8.png)
![S2-1440-light-diff-threshold8.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/fbd23b181cc42d1b/S2-1440-light-diff-threshold8.png)
![S2-1440-sheet.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/f9444e7f585789f7/S2-1440-sheet.png)
![S2-390-dark-diff-amp8.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/22189581cd6ca2ae/S2-390-dark-diff-amp8.png)
![S2-390-dark-diff-threshold8.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/b79cad8b7a297341/S2-390-dark-diff-threshold8.png)
![S2-390-light-diff-amp8.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/fab5cd27331fec21/S2-390-light-diff-amp8.png)
![S2-390-light-diff-threshold8.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/744af7ee5bb3214b/S2-390-light-diff-threshold8.png)
![S2-390-sheet.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/9ce5f90f28d4a4df/S2-390-sheet.png)
![S2-after-1440-dark.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/9929698aadc46a92/S2-after-1440-dark.png)
![S2-after-1440-light.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/04d0f37519227c21/S2-after-1440-light.png)
![S2-after-390-dark.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/55e2ace16c41296d/S2-after-390-dark.png)
![S2-after-390-light.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/8cc902ccdd67dfef/S2-after-390-light.png)
![S2-before-1440-dark.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/5afa45905f801693/S2-before-1440-dark.png)
![S2-before-1440-light.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/a8f211d3880e1ce5/S2-before-1440-light.png)
![S2-before-390-dark.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/644b51d43d50bf84/S2-before-390-dark.png)
![S2-before-390-light.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/2b448e0f92679eba/S2-before-390-light.png)
![S3-1440-dark-diff-amp8.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/7917f44cbf3d35f3/S3-1440-dark-diff-amp8.png)
![S3-1440-dark-diff-threshold8.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/623cc4057c3ca58f/S3-1440-dark-diff-threshold8.png)
![S3-1440-light-diff-amp8.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/241fea0e8e34644c/S3-1440-light-diff-amp8.png)
![S3-1440-light-diff-threshold8.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/5fb63463f3cea2e7/S3-1440-light-diff-threshold8.png)
![S3-1440-sheet.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/7f3c930f7e969ecc/S3-1440-sheet.png)
![S3-390-dark-diff-amp8.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/2b73555465d87f4c/S3-390-dark-diff-amp8.png)
![S3-390-dark-diff-threshold8.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/6bdc1d0e1ec21469/S3-390-dark-diff-threshold8.png)
![S3-390-light-diff-amp8.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/fda99b1c33e8d892/S3-390-light-diff-amp8.png)
![S3-390-light-diff-threshold8.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/fdd1fbc99d188d5a/S3-390-light-diff-threshold8.png)
![S3-390-sheet.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/66cefed27bc6b8fa/S3-390-sheet.png)
![S3-after-1440-dark.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/4e7af52488ac9526/S3-after-1440-dark.png)
![S3-after-1440-light.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/03b9633a3172da36/S3-after-1440-light.png)
![S3-after-390-dark.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/1a50e9461cf2d4ba/S3-after-390-dark.png)
![S3-after-390-light.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/c0e2d9abca7239a5/S3-after-390-light.png)
![S3-before-1440-dark.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/c8cb3cea532a7074/S3-before-1440-dark.png)
![S3-before-1440-light.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/1f22d1213d1bb410/S3-before-1440-light.png)
![S3-before-390-dark.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/c161a190df0eb2fd/S3-before-390-dark.png)
![S3-before-390-light.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/5814ed474132673d/S3-before-390-light.png)
[diff-statistics.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/30c1e670e2a8f78f/diff-statistics.json)
[parser-timing.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/80e68577f293b39b/parser-timing.json)
[published-snapshots.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/2f301ff46923bce2/published-snapshots.json)

[AFTER supplementary dismissal → reopen](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/692bac30d3df4ec4/R1-after-dismiss-reopen-1440-light.mp4)