## Evidence
**Second capture retake:** iPhone F02/F03 light and dark now use simultaneous before/after phones with matching, unedited native clock digits and zero changed status-bar pixels. F03 adds clear options frames. F07 now records the retained-query attempt on both iPhone revisions and the exact control boundary: Close clears the query before Filter returns. The direct query-retaining toggle remains supplied on Pixel. Native clocks remain live; no frozen native-clock override is claimed. The other evidence and the original artifact archive are preserved below.
Captured **8/8 required rows F01–F08**; seven are fully exercised on the primary iPhone light surface. F07 has an iPhone boundary recording plus the complete active-query toggle on Pixel. F02/F03 have paired recordings, native key stills and two diffs on **iPhone light, iPhone dark and Pixel light**. No S01–S09 supplementary fixture is claimed separately verified. This is evidence-only: no code changes, tests, commits, pushes, PR edits, CI checks or review verdict.
Source: [#27587](https://git.cafe/pingdotgg/t3code/pull/27587) ([GitHub mirror](https://github.com/pingdotgg/t3code/pull/27587)), before [32ecd717](https://github.com/pingdotgg/t3code/commit/32ecd71726fee16f335484c1f95cbce126d461f6), after [03ec49bf](https://github.com/pingdotgg/t3code/commit/03ec49bf7f1b7722a59db840a5ecb27c3fa94655). Exact full revisions are preserved in the rebuild guide and artifact metadata. Both JS revisions ran from their own pinned checkout with the same native fingerprint; the baseline checkout was removed afterward.
### Rendering and data
- **Every iPhone capture:** iPhone 17 Pro, iOS 27.2, portrait 402×874 pt, DPR 3, native 1206×2622 px, fontScale 1.0 (default Large), Expo dev client / React Native 0.88.0. Light/dark is stated in each row. Native screen recording and screenshot resolution are 1206×2622; no crop or registration.
- **Every Pixel capture:** Pixel 9, Android 16 / API 36, portrait 411.43×923.43 dp, DPR 2.625, native 1080×2424 px, fontScale 1.0, Expo dev client / React Native 0.88.0. All Pixel captures use light theme. Native recording and screenshot resolution are 1080×2424. Both Android F07 chunks are original native screenrecord outputs.
- **Clock:** app `Date.now = () => 1793610000000` (2026-11-02 09:00:00 UTC) on both revisions; this does not replace the Date constructor or freeze the OS clock. New iPhone stills use concurrent capture on two separately leased phones, with untouched and pixel-identical native status bars. No iOS screenshot normalization or persistent native time override was used.
- **Real app and providers:** normal Expo dev client, real server/RPC, real authenticated CLIProxyAPI-backed Claude 2.1.303 (`claudeAgent`, ready, installed, auth authenticated). Codex 0.163.0 is the second eligible provider. Claude has 13 models, including 9 legacy; target `claude-opus-4-8` / Claude Opus 4.8 is legacy. The sanitized before/after/retake catalog snapshots are identical. No provider-status fixture, static fixture, copied database or isolated component harness was used.
- **Seed B:** fresh isolated database at each revision's own migration level, `T3CODE_DEV_SEED=0`; synthetic Model search evidence project and unstarted thread. Current selection Claude Opus 5.5; its current-model favorite is on for F01–F04. Target Opus 4.8 starts unselected, unfavorited, unhidden. All providers; Show legacy off; Manage hidden off. F05 removes the current favorite through the recorded UI so the Favorites empty state is genuine. F06's favorite/hidden mutations and their reversal are also recorded. F08 starts with current GPT-6-Astra in a fresh unsaved scheduled-task editor.
[Before provider catalog](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/5ae26360217ab6fc/provider-catalog-before.json) · [After provider catalog](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/91e70ad04b9772f0/provider-catalog-after.json) · [Retake catalog](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/ac4cd7aa72910369/provider-catalog-retake.json) · [Retake preference restoration RPC](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/7b182e99d9863d24/retake-preferences.json) · [iOS environment](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/23c6b71c8fc1441a/ios-environment.json) · [Android environment](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/7f973ec74ae152df/android-environment.json) · [Native video metadata](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/31cf58be723a3bbc/video-metadata.json)
Every first-round retained still was selected only after two full screenshots with a 500 ms wait between screenshot commands compared pixel-identical. [Audit of 14 retained stills (12 primary, two superseded Pixel stills)](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/4411ed8b4178e626/settled-frame-audit.json) records dimensions and hashes. Every published still/diff and each recording's decoded review sheets were opened and visually inspected. Diff0 is absolute RGB difference amplified **32×**, including one-level changes; DiffT marks maximum-channel difference **>8/255** in magenta. Raw screenshots remain untouched; neither diff masks the status bar.
**Timing convention:** the new paired iPhone tables use pairlib.py command-dispatch times; the other times below are the uploaded driver's command timeline, written above each recording pair. `capture.py` timestamps command dispatch; `drive.py` timestamps command completion. Remote startup and encoding can shift playback positions, so use the labeled contact sheets for video positions. The same search sequence and catalog/preferences were used on both revisions; Pixel retake capture-v2 adds a return-to-top before the final clear because the longer after list scrolls the search field offscreen; remote command latency changes the cadence.
### F01 — blank and three-space reset
iPhone light new-task picker, seed B. In the after clip: clear at 00:03; focus, press Space at 00:08 / 00:10 / 00:12; clear at 00:14. Target stays omitted and default provider collapse/membership is restored. The shared before clip also contains this sequence; no separate F01 static comparison is claimed.
| F01 · iPhone 17 Pro · light | Before
[32ecd717](https://github.com/pingdotgg/t3code/commit/32ecd71726fee16f335484c1f95cbce126d461f6) | After
[03ec49bf](https://github.com/pingdotgg/t3code/commit/03ec49bf7f1b7722a59db840a5ecb27c3fa94655) | Diff0 | DiffT |
| --- | --- | --- | --- | --- |
| Combined search recordings | [before-iphone-light.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/0fce414331ab634d/before-iphone-light.mp4) | [after-iphone-light.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/009ae071d6cb3ccf/after-iphone-light.mp4) | No still pair | No still pair |
Environment: iPhone 17 Pro, iOS 27.2, portrait 402×874 pt, DPR 3, native 1206×2622 px, fontScale 1.0 (default Large), Expo dev client / React Native 0.88.0. Light. F01 uses the first segment of the original head recording linked above; the F02/F03 iPhone clips below are the new paired retakes.
### F02 — exact legacy search and normalization
The new paired iPhone recordings run `Opus 4.8` → dismiss keyboard / settled key frame → clear → `Claude Opus 4.8` → clear → ` cLaUdE oPuS 4.8 ` → clear → `claude-opus-4-8` → clear. Before has No matching models; after has one unselected Claude Opus 4.8 row with Legacy badge and empty star. Clearing removes it again. Same seed, preferences, selection and driver on both revisions; Show legacy remains off. Pixel evidence below is unchanged from round one.
**iPhone 17 Pro · light · simultaneous retake.**
Environment: iPhone 17 Pro, iOS 27.2, portrait 402×874 pt, DPR 3, native 1206×2622 px, fontScale 1.0 (default Large), Expo dev client / React Native 0.88.0. Two distinct simultaneously leased simulators on simbox-1; separate fresh server and Metro per revision. Theme: light.
| Query / action | Before driver dispatch | After driver dispatch |
| --- | --- | --- |
| `Opus 4.8` | 00:14.30 | 00:14.42 |
| `Claude Opus 4.8` | 00:29.81 | 00:30.43 |
| ` cLaUdE oPuS 4.8 ` | 00:44.34 | 00:44.33 |
| `claude-opus-4-8` | 00:56.57 | 00:57.25 |
| Before recording · [32ecd717](https://github.com/pingdotgg/t3code/commit/32ecd71726fee16f335484c1f95cbce126d461f6) | After recording · [03ec49bf](https://github.com/pingdotgg/t3code/commit/03ec49bf7f1b7722a59db840a5ecb27c3fa94655) |
| --- | --- |
| [iphone-light-before.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/7faddfd7768375fe/iphone-light-before.mp4) | [iphone-light-after.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/fba7cd7ede671409/iphone-light-after.mp4) |
[Driver command timeline](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/aa12bd7e3b52d1d7/steps-public.json) · [Before decoded review sheet](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/64b397887b742421/iphone-light-before-linear-sheet-01.png) · [After decoded review sheet](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/7ae727774a40d9c2/iphone-light-after-linear-sheet-01.png)
**iPhone 17 Pro · dark · simultaneous retake.**
Environment: iPhone 17 Pro, iOS 27.2, portrait 402×874 pt, DPR 3, native 1206×2622 px, fontScale 1.0 (default Large), Expo dev client / React Native 0.88.0. Two distinct simultaneously leased simulators on simbox-1; separate fresh server and Metro per revision. Theme: dark.
| Query / action | Before driver dispatch | After driver dispatch |
| --- | --- | --- |
| `Opus 4.8` | 00:09.30 | 00:09.91 |
| `Claude Opus 4.8` | 00:27.24 | 00:27.24 |
| ` cLaUdE oPuS 4.8 ` | 00:42.11 | 00:42.05 |
| `claude-opus-4-8` | 00:55.48 | 00:55.60 |
| Before recording · [32ecd717](https://github.com/pingdotgg/t3code/commit/32ecd71726fee16f335484c1f95cbce126d461f6) | After recording · [03ec49bf](https://github.com/pingdotgg/t3code/commit/03ec49bf7f1b7722a59db840a5ecb27c3fa94655) |
| --- | --- |
| [iphone-dark-before.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/7520c568d3a4812b/iphone-dark-before.mp4) | [iphone-dark-after.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/76e22256406ca6b9/iphone-dark-after.mp4) |
[Driver command timeline](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/3b96b5f5c0f9f453/steps-public.json) · [Before decoded review sheet](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/8d6c38fe619bb33e/iphone-dark-before-linear-sheet-01.png) · [After decoded review sheet](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/19cffdc34687afcd/iphone-dark-after-linear-sheet-01.png)
**Pixel 9 · light — steps and timestamps above the pair.**
| Query / action | Before driver time | After driver time |
| --- | --- | --- |
| `Opus 4.8` | 00:01.87 | 00:00.81 |
| `Claude Opus 4.8` | 00:13.24 | 00:22.44 |
| ` cLaUdE oPuS 4.8 ` | 00:16.63 | 00:27.28 |
| `claude-opus-4-8` | 00:19.74 | 00:30.44 |
| Before recording · [32ecd717](https://github.com/pingdotgg/t3code/commit/32ecd71726fee16f335484c1f95cbce126d461f6) | After recording · [03ec49bf](https://github.com/pingdotgg/t3code/commit/03ec49bf7f1b7722a59db840a5ecb27c3fa94655) |
| --- | --- |
| [before-pixel-light.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/074f1d498acf9bfb/before-pixel-light.mp4) | [after-pixel-light-retake.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/69f8a75bf10c5348/after-pixel-light-retake.mp4) |
Environment: Pixel 9, Android 16 / API 36, portrait 411.43×923.43 dp, DPR 2.625, native 1080×2424 px, fontScale 1.0, Expo dev client / React Native 0.88.0. Theme: light.
| F02 | Before
[32ecd717](https://github.com/pingdotgg/t3code/commit/32ecd71726fee16f335484c1f95cbce126d461f6) | After
[03ec49bf](https://github.com/pingdotgg/t3code/commit/03ec49bf7f1b7722a59db840a5ecb27c3fa94655) | Diff0 · zero threshold ×32 | DiffT · >8/255 |
| --- | --- | --- | --- | --- |
| iPhone 17 Pro · light |  |  |  |  |
| iPhone 17 Pro · dark |  |  |  |  |
| Pixel 9 · light |  |  |  |  |
Matching native clocks: light **10:04 / 10:04**; dark **10:07 / 10:07**. The entire unedited status bar has zero changed pixels in both new pairs. No time override, normalization, crop or mask was used.
### F03 — broad query, ordering, borders and scroll/reset
Continue the same new paired iPhone clips: clear, type `Claude`, dismiss keyboard, retain a top-anchored membership/ordering frame, scroll each list to its bottom anchor for an additional options frame, then return and clear. Before retains four current rows; after shows all 13 matching Claude rows. Selected/favorited Opus 5.5 stays first; nine Legacy rows are included and Haiku 4.5 is last. Both sides use the same top and bottom scroll anchors; the shorter before list already fits, so its bottom command reports no hidden content below. Absolute content offsets necessarily differ at the bottom of lists of different length.
**Use the bottom/options frame to judge Runtime and Full access:** all options and the Manage hidden switch sit fully above the search toolbar, with empty padding behind it. The complementary top frame shows the unchanged first rows and the inserted legacy rows; its lowermost options continue beneath the toolbar and are not the proof for those labels. The reset is recorded. The unchanged Pixel retake below includes its own return-to-top before clear.
**iPhone 17 Pro · light · simultaneous retake.**
Environment: iPhone 17 Pro, iOS 27.2, portrait 402×874 pt, DPR 3, native 1206×2622 px, fontScale 1.0 (default Large), Expo dev client / React Native 0.88.0. Two distinct simultaneously leased simulators on simbox-1; separate fresh server and Metro per revision. Theme: light.
| Query / action | Before driver dispatch | After driver dispatch |
| --- | --- | --- |
| `Claude` | 01:07.63 | 01:07.64 |
| Scroll to bottom / options | 01:21.36 | 01:20.36 |
| Before recording · [32ecd717](https://github.com/pingdotgg/t3code/commit/32ecd71726fee16f335484c1f95cbce126d461f6) | After recording · [03ec49bf](https://github.com/pingdotgg/t3code/commit/03ec49bf7f1b7722a59db840a5ecb27c3fa94655) |
| --- | --- |
| [iphone-light-before.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/7faddfd7768375fe/iphone-light-before.mp4) | [iphone-light-after.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/fba7cd7ede671409/iphone-light-after.mp4) |
[Driver command timeline](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/aa12bd7e3b52d1d7/steps-public.json) · [Before decoded review sheet](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/64b397887b742421/iphone-light-before-linear-sheet-01.png) · [After decoded review sheet](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/7ae727774a40d9c2/iphone-light-after-linear-sheet-01.png)
**iPhone 17 Pro · dark · simultaneous retake.**
Environment: iPhone 17 Pro, iOS 27.2, portrait 402×874 pt, DPR 3, native 1206×2622 px, fontScale 1.0 (default Large), Expo dev client / React Native 0.88.0. Two distinct simultaneously leased simulators on simbox-1; separate fresh server and Metro per revision. Theme: dark.
| Query / action | Before driver dispatch | After driver dispatch |
| --- | --- | --- |
| `Claude` | 01:07.27 | 01:06.39 |
| Scroll to bottom / options | 01:21.25 | 01:20.30 |
| Before recording · [32ecd717](https://github.com/pingdotgg/t3code/commit/32ecd71726fee16f335484c1f95cbce126d461f6) | After recording · [03ec49bf](https://github.com/pingdotgg/t3code/commit/03ec49bf7f1b7722a59db840a5ecb27c3fa94655) |
| --- | --- |
| [iphone-dark-before.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/7520c568d3a4812b/iphone-dark-before.mp4) | [iphone-dark-after.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/76e22256406ca6b9/iphone-dark-after.mp4) |
[Driver command timeline](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/3b96b5f5c0f9f453/steps-public.json) · [Before decoded review sheet](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/8d6c38fe619bb33e/iphone-dark-before-linear-sheet-01.png) · [After decoded review sheet](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/19cffdc34687afcd/iphone-dark-after-linear-sheet-01.png)
**Pixel 9 · light — broad-query step above the pair:** `Claude` at before 00:26.22, after 00:36.92; scroll to bottom at before 00:36.28, after 00:53.73.
| Before recording · [32ecd717](https://github.com/pingdotgg/t3code/commit/32ecd71726fee16f335484c1f95cbce126d461f6) | After recording · [03ec49bf](https://github.com/pingdotgg/t3code/commit/03ec49bf7f1b7722a59db840a5ecb27c3fa94655) |
| --- | --- |
| [before-pixel-light.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/074f1d498acf9bfb/before-pixel-light.mp4) | [after-pixel-light-retake.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/69f8a75bf10c5348/after-pixel-light-retake.mp4) |
Environment: Pixel 9, Android 16 / API 36, portrait 411.43×923.43 dp, DPR 2.625, native 1080×2424 px, fontScale 1.0, Expo dev client / React Native 0.88.0. Theme: light.
| F03 · complementary frames | Before
[32ecd717](https://github.com/pingdotgg/t3code/commit/32ecd71726fee16f335484c1f95cbce126d461f6) | After
[03ec49bf](https://github.com/pingdotgg/t3code/commit/03ec49bf7f1b7722a59db840a5ecb27c3fa94655) | Diff0 · zero threshold ×32 | DiffT · >8/255 |
| --- | --- | --- | --- | --- |
| iPhone 17 Pro · light · rows at top |  |  |  |  |
| iPhone 17 Pro · light · options at bottom |  |  |  |  |
| iPhone 17 Pro · dark · rows at top |  |  |  |  |
| iPhone 17 Pro · dark · options at bottom |  |  |  |  |
| Pixel 9 · light |  |  |  |  |
Matching native clocks in both rows and options frames: light **10:05 / 10:05**, dark **10:08 / 10:08**. All four new pairs have zero changed status-bar pixels.
### F04 — existing-thread provider filters
Existing unstarted thread caller. Driver times: 00:04 open composer model control; 00:34 select Codex; 00:42 exact query / 00:45 keyboard dismissed → No matching models; 00:57 select Claude; 01:04 exact query → single target; 01:25 All providers; 01:34 exact query → single target. The selected provider persists while typing and headers are non-collapsible. iOS search replaces Filter with Close, so filter changes visibly close/clear the prior search and retype the identical query; query persistence across the filter-menu action itself is not claimed.
| F04 · iPhone 17 Pro · light | Before
[32ecd717](https://github.com/pingdotgg/t3code/commit/32ecd71726fee16f335484c1f95cbce126d461f6) | After
[03ec49bf](https://github.com/pingdotgg/t3code/commit/03ec49bf7f1b7722a59db840a5ecb27c3fa94655) | Diff0 | DiffT |
| --- | --- | --- | --- | --- |
| After recording minimum | Not captured as a separate regression comparison | [F04-after-iphone-light.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/074d717af9cc127f/F04-after-iphone-light.mp4) | No still pair | No still pair |
Environment: iPhone 17 Pro, iOS 27.2, portrait 402×874 pt, DPR 3, native 1206×2622 px, fontScale 1.0 (default Large), Expo dev client / React Native 0.88.0. Light. Public step log: [F04 commands](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/63a516b5fd17c621/steps-public.json).
### F05 — Favorites cannot leak a non-favorite legacy model
New-task caller; same original combined head clip as F06 and the archived first-round F07. Driver times: 00:32 remove current-model favorite; 00:40 Favorites; 00:48 exact target query → no match; 01:12 All providers; 01:20 query; 01:24 star target; 01:34 Favorites; 01:42 query → target with filled star; 01:53 clear → target remains; 01:58 remove star → No favorite models; 02:08 nonempty query → No matching models. No favorite remains at this boundary.
| F05 · iPhone 17 Pro · light | Before
[32ecd717](https://github.com/pingdotgg/t3code/commit/32ecd71726fee16f335484c1f95cbce126d461f6) | After
[03ec49bf](https://github.com/pingdotgg/t3code/commit/03ec49bf7f1b7722a59db840a5ecb27c3fa94655) | Diff0 | DiffT |
| --- | --- | --- | --- | --- |
| After recording minimum | Not captured as a separate regression comparison | [F05-F07-after-iphone-light.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/f720afa6b3cf90e4/F05-F07-after-iphone-light.mp4) | No still pair | No still pair |
Environment: iPhone 17 Pro, iOS 27.2, portrait 402×874 pt, DPR 3, native 1206×2622 px, fontScale 1.0 (default Large), Expo dev client / React Native 0.88.0. Light.
### F06 — explicit hidden exclusion and manager restoration
Continue combined clip. Driver times: 02:22 All providers; 02:30 exact query → target; 02:34 Manage hidden on; 02:51 Hide eye → disabled target, Hidden marker and Show eye; 02:54 star hidden target; 02:57 manager off → no match. At 03:05 Show legacy on; 03:13 query still yields no match despite favorite plus explicit legacy bypass. 03:17 manager on; 03:19 Show eye; 03:22 manager off → target restored; 03:25 remove target favorite. Hidden and favorite state are reset before the original F07 segment.
| F06 · iPhone 17 Pro · light | Before
[32ecd717](https://github.com/pingdotgg/t3code/commit/32ecd71726fee16f335484c1f95cbce126d461f6) | After
[03ec49bf](https://github.com/pingdotgg/t3code/commit/03ec49bf7f1b7722a59db840a5ecb27c3fa94655) | Diff0 | DiffT |
| --- | --- | --- | --- | --- |
| After recording minimum | Not captured as a separate regression comparison | [F05-F07-after-iphone-light.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/f720afa6b3cf90e4/F05-F07-after-iphone-light.mp4) | No still pair | No still pair |
Environment: iPhone 17 Pro, iOS 27.2, portrait 402×874 pt, DPR 3, native 1206×2622 px, fontScale 1.0 (default Large), Expo dev client / React Native 0.88.0. Light.
### F07 — legacy toggle/search independence
**iPhone 17 Pro · light · recorded unavailability boundary on both revisions.** The direct Show legacy on → off action with the query retained has no reachable control on this iPhone layout. Both new recordings turn Show legacy on, enter `Opus 4.8`, dismiss the keyboard, tap the Thread settings header, then scroll bottom and top while keeping the query. The query remains visible; the native toolbar has the search field, Clear text and Close, but no Filter. Catalog exposes only Manage hidden models on iOS, without Android’s Legacy models alias. Pressing Close collapses search **and clears the query** before restoring Filter. The next menu visibly has Show legacy checked; toggling it off is therefore necessarily a blank-query action. Retyping the query afterward keeps the target only on the after revision. The query is then cleared and Filter returns. This documents why the retained-query interaction does not exist on iPhone; it does not claim it passed there.
Environment: iPhone 17 Pro, iOS 27.2, portrait 402×874 pt, DPR 3, native 1206×2622 px, fontScale 1.0 (default Large), Expo dev client / React Native 0.88.0. Two distinct simultaneously leased simulators on simbox-1; separate fresh server and Metro per revision. Theme: light. Same seeded preferences and current selection as F02/F03.
| Action | Before driver dispatch | After driver dispatch |
| --- | --- | --- |
| Turn Show legacy on | 00:11.71 | 00:11.71 |
| Enter `Opus 4.8` | 00:18.31 | 00:18.32 |
| Dismiss keyboard | 00:20.37 | 00:20.36 |
| Tap header with query retained | 00:26.53 | 00:26.46 |
| Scroll bottom with query retained | 00:32.09 | 00:32.09 |
| Scroll top with query retained | 00:37.52 | 00:37.51 |
| Close clears query / restores Filter | 00:45.08 | 00:45.08 |
| Open checked legacy menu | 00:47.79 | 00:47.79 |
| Turn Show legacy off (query is now blank) | 00:51.98 | 00:51.98 |
| Retype `Opus 4.8` | 00:58.20 | 00:58.20 |
| Before recording · [32ecd717](https://github.com/pingdotgg/t3code/commit/32ecd71726fee16f335484c1f95cbce126d461f6) | After recording · [03ec49bf](https://github.com/pingdotgg/t3code/commit/03ec49bf7f1b7722a59db840a5ecb27c3fa94655) |
| --- | --- |
| [F07-iphone-light-before.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/422a046c9790aad4/F07-iphone-light-before.mp4) | [F07-iphone-light-after.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/c7836855b4e0c4d1/F07-iphone-light-after.mp4) |
[Driver command timeline](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/c9030d401ee6ab44/steps-public.json) · [Before decoded review sheet](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/f5a57164196758ab/F07-iphone-light-before-linear-sheet-01.png) · [After decoded review sheet](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/f2a1034c56e33511/F07-iphone-light-after-linear-sheet-01.png)
[Accessibility snapshots: retained query, absent Filter, Close reset, checked menu](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/932ea23a0bfa1ce1/accessibility-evidence.json)
| F07 · iPhone 17 Pro · light | Before
[32ecd717](https://github.com/pingdotgg/t3code/commit/32ecd71726fee16f335484c1f95cbce126d461f6) | After
[03ec49bf](https://github.com/pingdotgg/t3code/commit/03ec49bf7f1b7722a59db840a5ecb27c3fa94655) | Diff0 · zero threshold ×32 | DiffT · >8/255 |
| --- | --- | --- | --- | --- |
| Legacy on + active query |  |  |  |  |
| Query retained after attempted alternate paths |  |  |  |  |
| Legacy off + query re-entered |  |  |  |  |
All three pairs show **10:10 / 10:10**, with zero changed native status-bar pixels. The first two pairs differ only at 198 pixels along the Legacy badge’s lower edge, maximum channel difference 6/255; DiffT (>8/255) has zero changed pixels. The last pair shows the expected search membership difference.
**Pixel completion of F07, same head and target preferences:** exact query is present while the filter menu's Show legacy check changes from on to off (driver 01:20), and the target remains. The Catalog → Legacy models alias is also exercised; after a covered-control attempt, keyboard dismissal exposes the switch, it goes off at 03:23 while target remains, and query clear at 03:27 restores the four-current-model catalog. The first switch tap after Android Back only dismissed the menu; it is not counted as a toggle. The split is the native 180-second screenrecord limit; both original chunks follow in order.
[F07-after-pixel-light.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/e010469d91f22b60/F07-after-pixel-light.mp4)
[F07-after-pixel-light.part-002.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/801a7968f5ce5bb9/F07-after-pixel-light.part-002.mp4)
Environment: Pixel 9, Android 16 / API 36, portrait 411.43×923.43 dp, DPR 2.625, native 1080×2424 px, fontScale 1.0, Expo dev client / React Native 0.88.0. Light. Public step log: [Pixel F07 commands](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/7a7de9e857eb8627/steps-public.json).
### F08 — scheduled-editor staging, clear, cancel/save and selection exception
Settings → Scheduled tasks → New task → synthetic project → Model. Driver times: 01:45 picker opens with current GPT-6-Astra; 02:09 search target; 02:13 select → check and Save; 02:16 clear / close → staged legacy remains. 02:27 Cancel returns to original current model; 02:30 reopen confirms it. Repeat search at 02:53, select at 02:57, **picker** Save at 03:00; editor now says Claude Opus 4.8. Reopen at 03:03 with blank query → selected legacy retained. The outer unsaved task form is discarded at 04:28; final list says No scheduled tasks yet. No task was created or run. This existing cross-provider selection also changes capability controls: Reasoning Medium becomes High and Service Tier is replaced by Fast Mode; this is part of the staged selection, not a before/after change attributed to this PR.
| F08 · iPhone 17 Pro · light | Before
[32ecd717](https://github.com/pingdotgg/t3code/commit/32ecd71726fee16f335484c1f95cbce126d461f6) | After
[03ec49bf](https://github.com/pingdotgg/t3code/commit/03ec49bf7f1b7722a59db840a5ecb27c3fa94655) | Diff0 | DiffT |
| --- | --- | --- | --- | --- |
| After recording minimum | Not captured as a separate regression comparison | [F08-after-iphone-light.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/441d4c28958f624e/F08-after-iphone-light.mp4) | No still pair | No still pair |
Environment: iPhone 17 Pro, iOS 27.2, portrait 402×874 pt, DPR 3, native 1206×2622 px, fontScale 1.0 (default Large), Expo dev client / React Native 0.88.0. Light. Public step log: [F08 commands](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/1ea4aca0e6eb03d2/steps-public.json).
### What else changed — new iPhone pairs
1. No status-bar change remains in any of the nine retaken pairs: light F02 10:04, light F03 10:05, dark F02 10:07, dark F03 10:08, F07 10:10. Each pair’s digits match; both sides are individually settled. Native clocks still advance in the original recordings.
2. F02 replaces the empty-result message/space with the Claude heading and single unselected Legacy row, empty favorite star and no check. Options/Catalog shift upward; Reasoning Medium, Fast Mode off, Full access and Manage hidden off stay unchanged.
3. F03 inserts nine Legacy rows. The selected/favorited first Opus 5.5 and the first current rows keep their order. Current Haiku 5.5 moves among the inserted rows; Haiku 4.5 becomes last. Group height, separators and final rounded border change; options move below the first viewport.
4. F03 uses matching top and bottom anchors, rather than forcing equal numeric offsets on unequal-length lists. The top frame proves ordering; the extra bottom frame proves all options without bottom-toolbar occlusion. Scrolled catalog content changes the native glass header’s backdrop; the two full-screen diffs retain that difference. No image is cropped, masked, aligned or registered.
5. F07 legacy-on and retained-query attempt pairs differ only at 198 pixels on the lower Legacy-badge edge (native x480–628, y572–587), maximum channel delta 6/255. Their >8/255 diffs are empty. This small rendering residual is preserved. The legacy-off/re-entered-query frame has the same expected membership and options displacement as F02.
6. Videos retain natural keyboard focus/dismissal, AutoFill bubbles, live native clocks and remote input cadence. Their dispatch timeline differs from native playback timing; use the decoded sheets for playback positions. They are original native MP4s, without cropping, retiming, combination or re-encoding.
No claim of zero differences outside the behavior is made. Every full-screen zero-threshold diff is available.
### Unverified and omitted
- **F07, iPhone active-query toggle:** recorded as unavailable on this native toolbar after keyboard dismissal, header tap and scrolling. Close clears the query; iOS Catalog lacks the legacy alias. Both revisions now have attempt recordings and stills. The direct toggle with query retained and Catalog alias remain covered on Pixel; no iPhone success is claimed for an interaction with no reachable control.
- **Fully frozen native clocks:** not claimed. App Date.now is fixed; native time is live. Simultaneous paired captures remove the clock difference from every new iPhone still and both diff thresholds.
- No separate baseline recordings for F04/F05/F06/F08 or static diffs for those after-only regression rows; their original after-recording minimum is preserved. F01 remains incidental in the original baseline combined clip. New F02/F03/F07 iPhone evidence has paired recordings, stills and both diff thresholds. No other rows were retested.
- S01–S09 remain supplementary inventory; no separate verification is claimed. CI, tests and external reviews were not re-run in this evidence-only round.
### Second-retake replay and audit
[Concurrent-phone rebuild guide](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/f8ead55a4397d844/rebuild-retake.md) · [Nine-pair settled-frame and native-clock audit](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/2d36b55cdef8f389/settled-clock-audit.json) · [Pair metrics](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/51cdaa56529527fd/pair-metrics.json) · [Six native video metadata records](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/5d3a9598699cb105/video-metadata.json) · [Before runtime environment](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/4b5e01003019ca4c/environment-before.json) · [After runtime environment](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/2b5151b9e787a604/environment-after.json) · [Before sanitized catalog](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/85b075f8b5bf5df0/provider-catalog-before.json) · [After sanitized catalog](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/9f29599f8b5ad873/provider-catalog-after.json)
All 18 retained screenshots, 18 diff images and six decoded recording sheets were opened and inspected before upload. Each retained screenshot also matches its two source captures; the helper waited 500 ms between completed screenshot rounds. Both sanitized catalogs compare exactly equal. Both source captures for each retained image are also uploaded below and match its SHA-256 exactly. Raw startup/pairing/config/session logs and disposable database homes stay private scratch; uploads contain synthetic evidence and sanitized metadata.
[Full retake gallery](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/d3979c2b6f953c1f/retake-gallery.html) · [F03 light options comparison](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/7cd3a473449f198c/F03-light-retake-summary.png)
**Exact duplicate stability sources:** each A/B source below is byte-identical to its retained screenshot; compare the SHA-256 in the audit.
| Run / frame / side | Source A | Source B |
| --- | --- | --- |
| 20261103-060427-iphone-light / F02-exact / before | [Capture A](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/90d606ac2605e3ea/F02-exact-before-settle-0-a.png) | [Capture B](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/beea482db8a0bbd1/F02-exact-before-settle-0-b.png) |
| 20261103-060427-iphone-light / F02-exact / after | [Capture A](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/b67ce72bdaa4016a/F02-exact-after-settle-0-a.png) | [Capture B](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/ed1521558587799e/F02-exact-after-settle-0-b.png) |
| 20261103-060427-iphone-light / F03-options / before | [Capture A](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/cc58f963550d8ae3/F03-options-before-settle-0-a.png) | [Capture B](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/130de117db056089/F03-options-before-settle-0-b.png) |
| 20261103-060427-iphone-light / F03-options / after | [Capture A](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/cb22c21284c890f7/F03-options-after-settle-0-a.png) | [Capture B](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/ba93cb6b89cb6313/F03-options-after-settle-0-b.png) |
| 20261103-060427-iphone-light / F03-rows / before | [Capture A](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/6bd3ef3037d5080c/F03-rows-before-settle-0-a.png) | [Capture B](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/35dc34a51332bb77/F03-rows-before-settle-0-b.png) |
| 20261103-060427-iphone-light / F03-rows / after | [Capture A](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/39bdffa70363f2f6/F03-rows-after-settle-0-a.png) | [Capture B](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/d9675187aebbca9b/F03-rows-after-settle-0-b.png) |
| 20261103-060649-iphone-dark / F02-exact / before | [Capture A](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/7008b1fdbac35f52/F02-exact-before-settle-0-a.png) | [Capture B](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/54cdc78ad9532a8f/F02-exact-before-settle-0-b.png) |
| 20261103-060649-iphone-dark / F02-exact / after | [Capture A](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/0777827db2c19408/F02-exact-after-settle-0-a.png) | [Capture B](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/c63849fe0f040807/F02-exact-after-settle-0-b.png) |
| 20261103-060649-iphone-dark / F03-options / before | [Capture A](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/f3d5742e161ce31c/F03-options-before-settle-0-a.png) | [Capture B](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/e89c5a943713c743/F03-options-before-settle-0-b.png) |
| 20261103-060649-iphone-dark / F03-options / after | [Capture A](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/9bdc07ca4b3cdb7b/F03-options-after-settle-0-a.png) | [Capture B](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/7156d6c48e15cfeb/F03-options-after-settle-0-b.png) |
| 20261103-060649-iphone-dark / F03-rows / before | [Capture A](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/9f32a77870366ccd/F03-rows-before-settle-0-a.png) | [Capture B](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/1260a443a152411e/F03-rows-before-settle-0-b.png) |
| 20261103-060649-iphone-dark / F03-rows / after | [Capture A](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/7a0bf4f1283719b1/F03-rows-after-settle-0-a.png) | [Capture B](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/b1a45c26f45da5ee/F03-rows-after-settle-0-b.png) |
| 20261103-060940-F07-iphone-light / F07-query-filter-unreachable / before | [Capture A](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/a3877f38078f1c0a/F07-query-filter-unreachable-before-settle-0-a.png) | [Capture B](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/dc2bcb21c74a4bee/F07-query-filter-unreachable-before-settle-0-b.png) |
| 20261103-060940-F07-iphone-light / F07-query-filter-unreachable / after | [Capture A](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/5a2f044d65004a34/F07-query-filter-unreachable-after-settle-0-a.png) | [Capture B](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/2beed1209f9e5f09/F07-query-filter-unreachable-after-settle-0-b.png) |
| 20261103-060940-F07-iphone-light / F07-query-legacy-off / before | [Capture A](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/a3443ace2344f7e8/F07-query-legacy-off-before-settle-0-a.png) | [Capture B](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/e0cf74371fd9138b/F07-query-legacy-off-before-settle-0-b.png) |
| 20261103-060940-F07-iphone-light / F07-query-legacy-off / after | [Capture A](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/1182bc162ba25e58/F07-query-legacy-off-after-settle-0-a.png) | [Capture B](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/f968fcf122056edc/F07-query-legacy-off-after-settle-0-b.png) |
| 20261103-060940-F07-iphone-light / F07-query-legacy-on / before | [Capture A](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/db248be5d138c2fd/F07-query-legacy-on-before-settle-0-a.png) | [Capture B](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/b3f166071485c2d2/F07-query-legacy-on-before-settle-0-b.png) |
| 20261103-060940-F07-iphone-light / F07-query-legacy-on / after | [Capture A](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/1f2c0b0690ade310/F07-query-legacy-on-after-settle-0-a.png) | [Capture B](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/66cfb826459e09d2/F07-query-legacy-on-after-settle-0-b.png) |
### Replay and scripts
[Rebuild guide](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/e424c6f2485b800b/rebuild.md) · [HTTP/RPC synthetic seed](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/9b25ccaa8d90d2b5/seed.mts) · [Original paired search/settled-frame driver](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/d9195e77d4ca3186/capture.py) · [Pixel retake driver with return-to-top](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/573109c47ccc1510/capture-v2.py) · [Regression recording driver](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/361846d9671cdea7/drive.py) · [Replay public command logs](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/ac6cd999303a8ec6/replay.py) · [Clock/environment inspector](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/f975ee7d7028949f/inspect.mjs) · [Diff generator](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/a886bf85b2469c06/diff.py) · [Three-second contact sheets](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/2ac995542d6c6bc3/sheets.py) · [Sequential paginated sheets](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/47fa33b1d6bbd20e/linear-sheets.py) · [Alternate seek-based sheet helper](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/bff94c529c9c8629/review-sheets.py) · [Settled-frame audit](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/95c55ffe603e77b5/audit-frames.py) · [Artifact packaging](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/3d7641c08df65eaf/package-artifacts.py) · [Original artifact uploader](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/6ac4f5c28a4dd696/upload-artifacts.py) · [Retake/document uploader](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/dd487961735f1211/upload-retake.py) · [Retake metadata/sheet packaging](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/c32220277bb5e1dd/retake-artifacts.py) · [Evidence markdown generator](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/623b0f35193af9ab/build-evidence.py).
Public step logs contain commands/times/exit codes only; private tool-session paths and raw account configuration were excluded. Every seed/driver/helper script written for this round is linked. Raw logs and rejected capture attempts remain in scratch and were not substituted for the retained evidence.
### Contact-sheet index
- [20261103-043537-before-iphone-light/sheet.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/77606e0d19511da7/sheet.png)
- [20261103-043737-before-iphone-dark/sheet.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/1c533da1037cf454/sheet.png)
- [20261103-043332-before-pixel-light/sheet.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/b01fc690185cac6c/sheet.png)
- [20261103-044410-after-iphone-light/sheet.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/ee5d19af3f364e15/sheet.png)
- [20261103-044201-after-iphone-dark/sheet.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/6f2bc8d61456d25d/sheet.png)
- [20261103-044206-after-pixel-light/sheet.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/e088dd1318f4d02e/sheet.png)
- [20261103-044900-F04-after-iphone-light/sheet.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/bd9821a007276b95/sheet.png)
- [20261103-045052-F05-F07-after-iphone-light/F05-F07-after-iphone-light-linear-sheet-01.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/790df964eb95474d/F05-F07-after-iphone-light-linear-sheet-01.png)
- [20261103-045052-F05-F07-after-iphone-light/F05-F07-after-iphone-light-linear-sheet-02.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/903d3f0f33c83564/F05-F07-after-iphone-light-linear-sheet-02.png)
- [20261103-045609-F08-after-iphone-light/F08-after-iphone-light-linear-sheet-01.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/e95f5bf473e3f27e/F08-after-iphone-light-linear-sheet-01.png)
- [20261103-045447-F07-after-pixel-light/F07-after-pixel-light-linear-sheet-01.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/8d352564398ddfea/F07-after-pixel-light-linear-sheet-01.png)
- [20261103-045447-F07-after-pixel-light/F07-after-pixel-light-linear-sheet-02.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/1499b395b8551ad4/F07-after-pixel-light-linear-sheet-02.png)
- [20261103-045447-F07-after-pixel-light/F07-after-pixel-light.part-002-linear-sheet-01.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/250c345d301c5244/F07-after-pixel-light.part-002-linear-sheet-01.png)
- [20261103-052445-after-pixel-light-retake/sheet.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/c2e4c56ecbd3b476/sheet.png)
### First-round complete upload output (verbatim; superseded iPhone assets are archived)
[steps-public.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/0c302373deb407f0/steps-public.json)
[before-iphone-light.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/0fce414331ab634d/before-iphone-light.mp4)
[steps-public.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/311038de0b9c8dc1/steps-public.json)
[before-iphone-dark.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/d6fc21a1457d346c/before-iphone-dark.mp4)
[steps-public.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/b6dd1d60378f8d14/steps-public.json)
[before-pixel-light.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/074f1d498acf9bfb/before-pixel-light.mp4)
[steps-public.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/8405e79d4560c2d6/steps-public.json)
[after-iphone-light.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/009ae071d6cb3ccf/after-iphone-light.mp4)
[steps-public.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/4e477b690ba51ea8/steps-public.json)
[after-iphone-dark.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/2d11a9d6ca95fffe/after-iphone-dark.mp4)
[steps-public.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/73e3c4a31aae04c2/steps-public.json)
[after-pixel-light.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/7e11e84f562af8f5/after-pixel-light.mp4)
[steps-public.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/63a516b5fd17c621/steps-public.json)
[F04-after-iphone-light.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/074d717af9cc127f/F04-after-iphone-light.mp4)
[steps-public.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/85c1055854a4663f/steps-public.json)
[F05-F07-after-iphone-light.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/f720afa6b3cf90e4/F05-F07-after-iphone-light.mp4)
[steps-public.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/1ea4aca0e6eb03d2/steps-public.json)
[F08-after-iphone-light.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/441d4c28958f624e/F08-after-iphone-light.mp4)
[steps-public.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/7a7de9e857eb8627/steps-public.json)
[F07-after-pixel-light.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/e010469d91f22b60/F07-after-pixel-light.mp4)
[F07-after-pixel-light.part-002.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/801a7968f5ce5bb9/F07-after-pixel-light.part-002.mp4)
[metrics.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/ce5d6aee0e293ce3/metrics.json)
[metrics.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/48f315eca4a48c93/metrics.json)
[metrics.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/056bb3dc323a7def/metrics.json)
[metrics.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/0c9719044515b4bc/metrics.json)
[metrics.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/70e5612b6d70a412/metrics.json)
[metrics.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/1e4da817c32cbb0f/metrics.json)
[seed.mts](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/9b25ccaa8d90d2b5/seed.mts)
[capture.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/d9195e77d4ca3186/capture.py)
[drive.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/361846d9671cdea7/drive.py)
[replay.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/ac6cd999303a8ec6/replay.py)
[inspect.mjs](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/f975ee7d7028949f/inspect.mjs)
[diff.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/a886bf85b2469c06/diff.py)
[sheets.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/2ac995542d6c6bc3/sheets.py)
[review-sheets.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/bff94c529c9c8629/review-sheets.py)
[package-artifacts.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/3d7641c08df65eaf/package-artifacts.py)
[provider-catalog-before.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/5ae26360217ab6fc/provider-catalog-before.json)
[provider-catalog-after.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/91e70ad04b9772f0/provider-catalog-after.json)
[ios-environment.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/23c6b71c8fc1441a/ios-environment.json)
[android-environment.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/7f973ec74ae152df/android-environment.json)
[linear-sheets.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/47fa33b1d6bbd20e/linear-sheets.py)
[audit-frames.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/95c55ffe603e77b5/audit-frames.py)
[settled-frame-audit.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/058b40eea1978d08/settled-frame-audit.json)
[rebuild.md](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/101b4fe47adda55b/rebuild.md)
[upload-artifacts.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/6ac4f5c28a4dd696/upload-artifacts.py)
[video-metadata.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/74eb85c79833966e/video-metadata.json)
[build-evidence.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/c50e39210245277b/build-evidence.py)
[artifact-files.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/a354666abfd8b48c/artifact-files.json)
[uploaded-artifacts.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/4b68235b76634625/uploaded-artifacts.json)
[uploads-verbatim.md](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/4b761d300e59e06d/uploads-verbatim.md)
[after-pixel-light-retake.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/69f8a75bf10c5348/after-pixel-light-retake.mp4)
[steps-public.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/40d9a057425bd45a/steps-public.json)
[metrics.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/bc791a2670421947/metrics.json)
[metrics.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/1dd9b788ff9cdc06/metrics.json)
[capture-v2.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/573109c47ccc1510/capture-v2.py)
[upload-retake.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/dd487961735f1211/upload-retake.py)
[retake-artifacts.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/c32220277bb5e1dd/retake-artifacts.py)
[provider-catalog-retake.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/ac4cd7aa72910369/provider-catalog-retake.json)
[retake-preferences.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/7b182e99d9863d24/retake-preferences.json)
[video-metadata.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/31cf58be723a3bbc/video-metadata.json)
[settled-frame-audit.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/4411ed8b4178e626/settled-frame-audit.json)
[artifact-files.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/4358576d915a35e4/artifact-files.json)
[rebuild.md](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/1d636ffe5ebe93b3/rebuild.md)
[build-evidence.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/65a141652c434a57/build-evidence.py)
[rebuild.md](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/e424c6f2485b800b/rebuild.md)
[build-evidence.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/623b0f35193af9ab/build-evidence.py)
### First-round cleanup and friction (preserved)
Server and Metro were stopped through their tracked execution sessions. All four Pixel sessions and the iPhone lease were closed. The baseline worktree was removed. The lead-owned after checkout remains clean at the pinned head. Disk free: 19 GB. No config outside the worktrees/scratch was edited.
FRICTION:
- tool | whitespace-only fill was rejected by the device proxy | permit literal whitespace arguments; real software-keyboard Space taps covered F01.
- tool | iOS active search replaces Filter with Close and clears query on close | keep a filter control available during search, or use Pixel for the direct-toggle substep.
- tool | iOS status-bar normalization caused a transient charging-icon pixel change | a persistent native status-bar override would permit a frozen OS clock without that transient.
- tool | Android search scrolled offscreen and screenrecord split after 180 seconds | the retake returns to top before clearing; all native recording chunks are linked.
- tool | one newly leased Pixel carried another session’s active recording marker | reset recording state when returning devices; that session was untouched and a fresh lease recorded successfully.
### Second-retake upload output (verbatim)
[metrics.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/14255bd54628e3c9/metrics.json)
[metrics.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/2fcad47722f7c9bf/metrics.json)
[metrics.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/4fa2c313f2a2953c/metrics.json)
[iphone-light-after.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/fba7cd7ede671409/iphone-light-after.mp4)
[iphone-light-before.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/7faddfd7768375fe/iphone-light-before.mp4)
[steps-public.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/aa12bd7e3b52d1d7/steps-public.json)
[metrics.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/20e658b76bb49407/metrics.json)
[metrics.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/579c50a4f6fa6552/metrics.json)
[metrics.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/f1fdefabdb10671f/metrics.json)
[iphone-dark-after.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/76e22256406ca6b9/iphone-dark-after.mp4)
[iphone-dark-before.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/7520c568d3a4812b/iphone-dark-before.mp4)
[steps-public.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/3b96b5f5c0f9f453/steps-public.json)
[metrics.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/741e9b18d6ea23a0/metrics.json)
[metrics.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/351c6fd6526745b5/metrics.json)
[metrics.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/c7a16e159b1fd6c6/metrics.json)
[F07-iphone-light-after.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/c7836855b4e0c4d1/F07-iphone-light-after.mp4)
[F07-iphone-light-before.mp4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/422a046c9790aad4/F07-iphone-light-before.mp4)
[steps-public.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/c9030d401ee6ab44/steps-public.json)
[accessibility-evidence.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/932ea23a0bfa1ce1/accessibility-evidence.json)
[seed.mts](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/c108b145fe8a23ad/seed.mts)
[capture.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/021671b2a53cb2bb/capture.py)
[drive.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/30353037055ed342/drive.py)
[inspect.mjs](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/a38397234c80cf09/inspect.mjs)
[diff.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/6bc98a69f0a54b5a/diff.py)
[linear-sheets.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/a81c311010f6cac9/linear-sheets.py)
[paired.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/01e2a41cf8c2473f/paired.py)
[pairlib.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/11bd171a362982d3/pairlib.py)
[f07.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/555cf2e47968a446/f07.py)
[both-command.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/06396b346f8725a1/both-command.py)
[audit.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/9c7ca244297e25e0/audit.py)
[package.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/9067ace227e839f3/package.py)
[upload.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/45e9a02bfb2f5617/upload.py)
[build-evidence.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/0f43f9310ddd5586/build-evidence.py)
[provider-catalog-before.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/85b075f8b5bf5df0/provider-catalog-before.json)
[provider-catalog-after.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/9f29599f8b5ad873/provider-catalog-after.json)
[environment-before.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/4b5e01003019ca4c/environment-before.json)
[environment-after.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/2b5151b9e787a604/environment-after.json)
[settled-clock-audit.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/2d36b55cdef8f389/settled-clock-audit.json)
[pair-metrics.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/51cdaa56529527fd/pair-metrics.json)
[video-metadata.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/5d3a9598699cb105/video-metadata.json)
[stopped-processes.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/8bebc1079094abdc/stopped-processes.json)
[run-notes.md](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/f3e3100d505450ad/run-notes.md)
[rebuild-retake.md](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/feb01e6080ca2209/rebuild-retake.md)
[package.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/68bb4e9059889a4e/package.py)
[build-evidence.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/3d90aa55489e3528/build-evidence.py)
[rebuild-retake.md](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/f8ead55a4397d844/rebuild-retake.md)
[run-notes.md](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/e33b59615cafe6bb/run-notes.md)
[gallery.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/936d575bf5f3196c/gallery.py)
[retake-gallery.html](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/d3979c2b6f953c1f/retake-gallery.html)
[upload-additions.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/17532a91f5c0f2f2/upload-additions.py)
[upload-stability.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/a0c7dee3d76bf748/upload-stability.py)
[build-evidence.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/28e5ca8b11f9e6e0/build-evidence.py)
### Second-retake cleanup and friction
Both leases closed; both backends and Metros stopped using confirmed port-owner PIDs; the base worktree was removed. The four ports have no listeners. The after checkout is clean at [03ec49bf](https://github.com/pingdotgg/t3code/commit/03ec49bf7f1b7722a59db840a5ecb27c3fa94655). Disk free: 19 GB. No code changes, commits, pushes, PR edits, tests or CI reruns.
FRICTION:
- runbook | opening sessions from main made worktree installs fail the workspace ownership check | open and drive each session from its owning worktree.
- tool | OS deep-link acceptance returned deadline/main-thread errors and left a redundant prompt although the app loaded and paired | reliable launch confirmation; Cancel dismissed the redundant prompt.
- tool | active iPhone search replaces Filter with Close, which clears the query | expose Filter during search if the direct retained-query interaction should be supported on iPhone.