# Current-build capture index: F1 and B1

Captured 2026-10-24 for [#4589](https://github.com/pingdotgg/t3code/issues/4589) / [#24400](https://github.com/pingdotgg/t3code/pull/24400). BEFORE [b1622197fa](https://git.cafe/pingdotgg/t3code/commit/b1622197fa0d2177053b57198b7bbc9e753b4f58); AFTER [86f313103d](https://git.cafe/pingdotgg/t3code/commit/86f313103db553087566fef071a1835a54c3ceb5). **2/2 requested new rows captured. Fresh independent proof audit remains the lead’s next step.**

[Execution matrix and original detailed requirements](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/8a4031a5d629c33e/matrix.md)

## Scope and runtime

Both real web clients used CSS 1440×900, DPR 2, light theme, loaded fonts, matching selected empty persisted stopped thread, sidebar width and closed details/terminal/right panels. PNG exports are 1280×800; full recordings are 2880×1800. Neither is mislabeled as a 1440×900 bitmap. No provider was launched. No source edits, tests, commits, pushes, public comments, server starts/stops, or live database writes were performed.

Actual browser fetch of the Vite client transform returned 43,256 characters with `streamFailureRetries` and `resubscribesOnSameSession` on AFTER, versus 40,606 with neither on BEFORE. Both subscribed server-config snapshots advertised `threadFind:true` and `threadFindProgressive:true`; the actual Find requests used `orchestration.searchThreadStream`. [runtime-proof.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/9a7831eab3bc1ff2/runtime-proof.json).

Natural incident trigger remains unknown. These recordings use **controlled invalid Chunk values at the RPC decoder boundary**, not React/atom mocks. F1 intercepts `MessageEvent.data` before decoding the first real matching search Chunk, installed before typing; it does not dispatch a late synthetic failure. The original real Chunk is retained in the transcript. The fault stays armed for matching search subscriptions until explicitly disarmed. B1 dispatches an invalid Chunk into the tracked current shell request only, with the app socket OPEN and genuine first values received before every fault.

## F1 — finite Find failure, manual Retry, completion and close/reopen

Normal control query `normal-control` completed with zero matches on both builds. Then `fault-query` was entered with receive interception armed. Two immediate query subscriptions appeared on each build; the active one is BEFORE request 87 / AFTER 76. Both initial real Chunk reads were intercepted. BEFORE automatically replaced request 87 with 89, 90 and 91, and stayed Searching. AFTER displayed Search failed / Retry and made no later automatic replacement for about 9.7 seconds. Its socket stayed OPEN. Injection was disarmed before clicking Retry; fresh request 78 completed with zero matches, then Close find → Ctrl+F → query completed again. BEFORE was disarmed and recovered through close/reopen.

**Narration / video times:** BEFORE normal query 00:36.3, fault query 00:41.9, retries 00:42.9 / 00:45.0 / 00:49.0, close/reopen query 00:57.3. AFTER normal query 00:20.1, fault query 00:25.7, manual Retry 00:35.7, close/reopen query 00:41.6. Initial idle footage includes a locator retry; it is not an app search delay.

**Recordings:** [BEFORE recording](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/0933dfcdb2edaaa7/F1-before.mp4) · [AFTER recording](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/2bcecdf31c54f722/F1-after.mp4).

| BEFORE [b1622197fa](https://git.cafe/pingdotgg/t3code/commit/b1622197fa0d2177053b57198b7bbc9e753b4f58) | AFTER [86f313103d](https://git.cafe/pingdotgg/t3code/commit/86f313103db553087566fef071a1835a54c3ceb5) | Diff0 ×8 | Diff >16 |
|---|---|---|---|
| ![BEFORE: Find keeps Searching during automatic retries](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/ee2a6fc940253822/F1-before.png) | ![AFTER: Search failed with manual Retry](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/34c55f6a6292e5a0/F1-after.png) | ![F1 RGB difference amplified eight times](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/a2fab5ce49b2a439/F1-diff0x8.png) | ![F1 per-channel difference above 16](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/36b9deb1b5bcf13d/F1-diff16.png) |

**What else changed:** 684 RGB-different pixels (552 with any channel >16), all inside exported rectangle `[1115,61,1195,85)`: Searching is replaced by the wrapped Search failed label and Retry control. No other pixels differ. Both independently repeated error endpoints are exact RGB matches, separated by 6,000ms BEFORE and 5,944ms AFTER: [BEFORE duplicate](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/7150e70c862727e7/F1-before-duplicate.png) / [AFTER duplicate](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/1ebf76da58d89d8c/F1-after-duplicate.png).

Completed zero-match close/reopen endpoints (pixel-identical across builds):

| BEFORE completed control | AFTER completed control |
|---|---|
| ![BEFORE: completed zero-match search after close/reopen](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/04ab07cfc61eb199/F1-complete-before.png) | ![AFTER: completed zero-match search after manual retry and close/reopen](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/760941fef9f21497/F1-complete-after.png) |

## B1 — repeated shell faults, healthy gap, cap and session replacement

Each fault targeted the current `orchestration.subscribeShell` request after genuine snapshot/synchronized values. The first nine retries stayed on BEFORE socket 5 and AFTER socket 4. A 3s healthy gap after retry 2 included successful unsettle/settle commands and real shell updates; the next AFTER delay remained 1s. Transcript timing measures fault → replacement Request, not request → request. Browser scheduling tolerance is visible in measured values.

**Narration / video times:** use the fault→replacement times below. Each recovery sends a real settle/unsettle command with successful receipt and visible sidebar/composer change. After the ninth recovery, a tenth fault is injected; only the app socket is closed 1s later. BEFORE has already retried at base delay, while AFTER has a pending capped retry. The new socket receives valid shell values, is faulted once, and returns to base delay. Both are observed past the old fault +32s, followed by successful normal unsettle → settle → unsettle. At about 02:24 both receive an additional matched settle/unsettle for final stills. Long static waits remain in the recording; the transcript is the primary timing evidence.

**Recordings:** [BEFORE recording](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/a546549985157b81/B1-before.mp4) · [AFTER recording](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/0265913f3700df29/B1-after.mp4).

| Fault | BEFORE measured ms / video fault→request | AFTER measured ms / video fault→request | Expected AFTER ms |
|---|---|---|---|
| 1 | 253.9 / 00:00.0→00:00.3 | 254.4 / 00:00.3→00:00.6 | 250 |
| 2 | 251.9 / 00:00.7→00:00.9 | 502.6 / 00:01.0→00:01.5 | 500 |
| 3 | 252.4 / 00:04.3→00:04.6 | 1002.4 / 00:04.9→00:05.9 | 1000 |
| 4 | 251.9 / 00:05.0→00:05.2 | 2002.3 / 00:06.3→00:08.3 | 2000 |
| 5 | 252.1 / 00:05.6→00:05.8 | 4004.4 / 00:08.7→00:12.7 | 4000 |
| 6 | 252.0 / 00:06.2→00:06.5 | 8003.2 / 00:13.1→00:21.1 | 8000 |
| 7 | 252.6 / 00:06.8→00:07.1 | 16012.4 / 00:21.5→00:37.5 | 16000 |
| 8 | 251.9 / 00:07.5→00:07.7 | 30007.0 / 00:37.9→01:07.9 | 30000 |
| 9 | 255.4 / 00:08.1→00:08.3 | 30003.1 / 01:08.3→01:38.3 | 30000 |

BEFORE: tenth fault 00:08.7, socket close 00:09.7; new socket 6, request 194→196, 252.0ms at 00:14.4→00:14.7. Old socket emitted **zero** shell Requests after close through 00:40.7 (old fault +32.0s).

AFTER: tenth fault 01:38.7, socket close 01:39.7; new socket 5, request 147→150, 251.8ms at 01:42.7→01:42.9. Old socket emitted **zero** shell Requests after close through 02:10.7 (old fault +32.0s).

All accepted-retake dispatch receipts succeeded (21 BEFORE / 25 AFTER, including app-generated receipts); no failed command is part of this retake. Eleven faults per build have genuine first values recorded before injection. Vite sockets were never closed.

**Narration for the endpoint pair:** final normal recovery is complete and both clients show the same unsettled empty stopped thread. Static stills prove matching presentation and recovery, not retry timing.

| BEFORE [b1622197fa](https://git.cafe/pingdotgg/t3code/commit/b1622197fa0d2177053b57198b7bbc9e753b4f58) | AFTER [86f313103d](https://git.cafe/pingdotgg/t3code/commit/86f313103db553087566fef071a1835a54c3ceb5) | Diff0 ×8 | Diff >16 |
|---|---|---|---|
| ![BEFORE: final normal unsettled thread after recovery](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/6e727b1112b6bc1f/B1-before.png) | ![AFTER: final normal unsettled thread after capped retries and session replacement](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/2222f4dcf142de93/B1-after.png) | ![B1 final RGB difference amplified eight times: zero pixels](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/4da6cb7d0193e022/B1-diff0x8.png) | ![B1 final threshold-16 difference: zero pixels](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/36c3a869b7644aa1/B1-diff16.png) |

**What else changed:** zero pixels across the retained final before/after endpoints. Both independent duplicate pairs are exact RGB matches, separated by 18,021ms BEFORE / 18,009ms AFTER: [BEFORE duplicate](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/91e74caa2c9ab18c/B1-before-duplicate.png) / [AFTER duplicate](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/d395e36019c16d92/B1-after-duplicate.png). An earlier pair had a fluctuating one-pixel sidebar edge (maximum channel delta 2) and was discarded; it was not accepted using a threshold or masked. [Retained timestamp metadata](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/09635ae9e3f946af/B1-stability-times.json) · [RGB report](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/05b09c563982c7d2/pixel-report.json).

## Reproduction, transcripts and inspection

Use the existing matching persisted stopped thread; no provider or DB seed is required. The lead’s copied HEAD project was retained; BASE’s extra automatically created project was deleted through normal `projects.mutate` with `force:true`. The shared thread was renamed through normal `thread.metadata.update`. Pair privately in separate profiles because localhost cookies span ports. Install the tracker, use the app-generated selected-thread route, close details/terminal/right panels, set light/1440×900, and wait for fonts. For F1 press Ctrl+F, run normal-control, arm the search interceptor before typing fault-query, retain the failure state for >1s, disarm, click Retry, close and reopen. For B1 apply the exact runtime patches, replace only the tracked app socket once for a fresh initial session, then execute the B1 driver. It waits for genuine first values, performs nine recoveries, a healthy gap, the pending-cap session replacement, old-deadline check and final normal actions.

Artifacts are sanitized: no pairing tokens, cookies, auth headers, settings, database, keys or authenticated socket URLs are uploaded. RPC request exports omit headers and socket URLs. Only relevant orchestration/search/project records are included; full valid frames for those records remain in the transcripts.

Primary transcripts: [BEFORE timed RPC transcript](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/b2110e294a34398e/before-transcript.json) · [AFTER timed RPC transcript](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/b58abb8a48f5c80d/after-transcript.json) · [recording/time-origin mapping](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/cd3fd5fc2581bde5/recording-clocks.json).

Current scripts: [driver.js](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/a5723e7bceef6566/driver.js) · [runtime-patches.js](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/7ee618607c167bbc/runtime-patches.js) · [search-arm.js](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/a6750ba736ace3f6/search-arm.js) · [b1-driver.js](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/5698beed3c92f0ed/b1-driver.js) · [fixture-commands.js](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/cb3996539722151d/fixture-commands.js) · [export-transcript.js](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/bf2148c6a94bc655/export-transcript.js) · [process-images.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/09e50c8b195f51e9/process-images.py) · [inspect.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/8b53f6402a1d0765/inspect.py) · [contact-pages.py](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/4063d834766fc1a0/contact-pages.py). The export was transferred in string slices because the preview tool limits responses to 64KB; no frames were dropped from the selected transcript.

Setup-only reference artifacts, not accepted B1 proof: [original-driver.js](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/7f1ad0f5de863a49/original-driver.js) · [before-transcript-initial.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/fcf4433d68f230d7/before-transcript-initial.json) · [after-transcript-initial.json](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/16d3096d1a1663f0/after-transcript-initial.json). Two B1 rehearsals were discarded (string request ID ignored; then missing unsettle reason); the accepted retake uses numeric request IDs and `reason:"user"` with successful receipts. The published initial setup transcripts are labeled separately.

All retained stills and both diffs per row were opened; duplicates were compared in RGB across every channel. All four recordings were inspected through one-second frames plus final frames. Contact sheets: [F1-images-contact.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/01251f9fa1a09cce/F1-images-contact.png) · [F1-before-video-contact.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/751b2092696fa035/F1-before-video-contact.png) · [F1-after-video-contact.png](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/8f209a8539b14eef/F1-after-video-contact.png).

B1 one-second inspection pages: [before 1](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/a27e0fe74b43f172/B1-before-contact-1.png) · [before 2](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/c6b9b22cf8327ea6/B1-before-contact-2.png) · [before 3](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/b4e9c2234a0974f4/B1-before-contact-3.png) · [before 4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/8a5d55974cf929bc/B1-before-contact-4.png) · [before 5](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/214d6dc344e64610/B1-before-contact-5.png) · [after 1](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/72977616e7ea2331/B1-after-contact-1.png) · [after 2](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/6e7fbcefecffec5c/B1-after-contact-2.png) · [after 3](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/8d0f4ee6760a88a3/B1-after-contact-3.png) · [after 4](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/a4eb213ae4f3c716/B1-after-contact-4.png) · [after 5](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/ebd9407b77f04912/B1-after-contact-5.png).

## Scope limits and FRICTION

FRICTION: the natural failure trigger is unknown; all decoder faults here are controlled and disclosed. CSS animations/transitions and carets were disabled equally; reduced-motion emulation was unavailable. This is recorded in the runtime metadata. PNG export downscales the real CSS viewport; recordings retain full DPR dimensions. Browser scheduling adds small timing overhead.

FRICTION: original V1–V6 evidence remains accepted **historical** execution at BASE [27e0c684c5](https://git.cafe/pingdotgg/t3code/commit/27e0c684c5) / HEAD [383a473f2e](https://git.cafe/pingdotgg/t3code/commit/383a473f2e), not current-HEAD evidence. [Historical index](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/e472b85181cef193/index.md). This capturer did not independently re-audit those six rows.

Unverified here: native desktop/mobile execution, responsive/dark refresh, remote/relay/tunnel, other provider/product-specific consumers, finite transport failure control, and the full test suite. No new four-way sweep was requested or performed. Lead-reported focused tests are not capturer-run tests. Lead reports a successful separate-tab integrated pass: real search Chunk request 26 failed, remained failed for about 5s without automatic retry, and manual Retry request 27 completed 0/0 on the same OPEN socket 0. Fresh independent proof audit remains pending.

Browser capture was explicitly released to the lead. Both isolated servers, sources and scratch remain intact for audit. No PR comments or edits were made.

[Index generation script](https://gh-file-drop-api-prod-mi5fy3sowv63ufte.pinglabs.workers.dev/f/0a12c3ec15c52857/build-index.py).