Findings: none remaining in the revised W/B narration for [#33283](https://git.cafe/pingdotgg/t3code/pull/33283), head [68414fb0aa](https://git.cafe/pingdotgg/t3code/commit/68414fb0aa3321c60eee6f714c4c5b586ed82b5e).

All three requested text corrections are present and supported by the recordings:

1. W after explicitly discloses the second read's “Loading output…” at 0.4 s and populated output by 0.5 s. Exact frame samples confirm both states.
2. B's 0.2 s before / 0.3 s after checkpoints now describe dispatch to the inactive thread, rather than a completed failure. The failure indicators appear by 1.4 s before / 2.2 s after, following blue working indicators at 1.3 s / 2.1 s.
3. The paragraph explicitly supersedes the W/B index's earlier 1.53 s / 2.68 s “settled” notification times. Its replacement holds at about 1.9 s / 2.9 s agree with the recordings, and entry/exit motion is disclosed.

No narrated W/B checkpoint contradicts the sampled frames, and no additional substantive visible transient was found that needs disclosure. No recapture is required.

| Row | Status | Recording checks |
|---|---|---|
| W | SUFFICIENT | Before: expanded at 0.4 s, collapsed at 5.2 s, no Resume control at 9.4 s. After: expanded at 0.4 s with the disclosed loading placeholder, populated at 0.5 s, collapsed at 4.9 s, Resume pending spinner at 9.2 s, prior-error notice visible at 9.5–9.7 s and absent at 9.8 s, successful reply and “Read 2 files” by 10.2 s. |
| B | SUFFICIENT | Before/after respectively: dispatch checkpoints 0.2/0.3 s; failure indicators by 1.4/2.2 s; notification holds about 1.9/2.9 s; dismissal in progress at 3.8/5.6 s; clean closed-folder views by 4.1/5.9 s; folder/hover interaction at 6.4/9.3 s; settled hover cards at 8.8/12.0 s. Approximately 0.5 s toast entry/exit motion is disclosed. |

Method and scope: read `data.resource.description` from `cafe pr view 33283 --repo pingdotgg/t3code --json`, concentrating on “Evidence added on the merged head” and its “Recording steps” paragraph. Downloaded the four original clips with curl and sampled exact frames with ffmpeg, including adjacent frames around expansion, failure, notification movement, Resume and completion. All four clips decode at 1440×1000 and 10 fps; durations are W before 11.5 s, W after 14.9 s, B before 10.7 s and B after 13.9 s. Timestamped inspection sheets were local scratch material only. A final PR read confirmed the same head and unchanged description (resource version 12).

Recordings checked: W [before](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/52722224e5583f23/W-flow-before.webm) / [after](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/9e8025f257545228/W-flow-after.webm); B [before](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/1b4c9ede0fbcf249/B-flow-before.webm) / [after](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/a27d7f571aba16a4/B-flow-after.webm). The [previous audit](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/3394001f6b38a93d/proof-audit-4.md) supplied the three correction requests and prior acceptances.

Stills, twins, masks and the other 14 accepted rows were not reopened. Their accepted status and existing limitations stand. This is a read-only narration/disclosure audit: no repository edits, pushes, PR comments, browser sessions or dev servers.

FRICTION: none

VERDICT: SUFFICIENT
