repremium/mokosh
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Milestone v1 (v2.0.0): Mokosh — Session Capture #1

Merged
strategy155 merged 297 commits from gsd/phase-04-harden-clean-up-optional into main 2026-05-31 15:34:17 +00:00
Conversation 0 Commits 297 Files Changed 235 +399 -179
1 changed files with 399 additions and 179 deletions
Download Patch File Download Diff File Expand all files Collapse all files
Showing only changes of commit 92810a0f34 - Show all commits

578
.planning/phases/01-stabilize-video-pipeline/01-REVIEW-FIX.md
View File

@@ -1,182 +1,37 @@
---
phase: 01-stabilize-video-pipeline
fixed_at: 2026-05-16T07:25:00Z
fixed_at: 2026-05-16T11:05:00Z
review_source: .planning/phases/01-stabilize-video-pipeline/01-REVIEW.md
iteration: 1
iteration: 2
findings_in_scope: 18
fixed: 5
skipped: 0
remaining: 13
status: partial
checkpoint: true
checkpoint_reason: context_budget_exhausted_after_critical_cluster
fixed: 16
skipped: 2
status: all_fixed
prior_partial_commit: 2e3f524
commit_range: 2e3f524..HEAD
---
# Phase 1: Code Review Fix Report (PARTIAL — CHECKPOINT)
# Phase 1: Code Review Fix Report (FINAL)
**Fixed at:** 2026-05-16T07:25:00Z
**Fixed at:** 2026-05-16T11:05:00Z
**Source review:** [01-REVIEW.md](./01-REVIEW.md)
**Iteration:** 1
**Status:** partial — CHECKPOINT for orchestrator continuation
**Iteration:** 2 (continuation of partial iteration 1)
**Status:** all_fixed
## Summary
- Findings in scope: 18 (3 Critical + 9 Warning + 6 Info)
- Fixed this pass: **5** (3 Critical + 2 Warning, all delivered as one
cohesive commit because they share the `encodeAndSendBuffer` code path)
- Skipped: 0
- Remaining: **13** (7 Warning + 6 Info) — see "Remaining Work" below
- Sweep items added: 0 (no spare context budget for the defensive sweep
on this pass — see "Sweep Status" below)
- Findings in scope: **18** (3 Critical + 9 Warning + 6 Info)
- Fixed: **16**
- Skipped (pre-approved documented skips, NOT context cuts): **2** (WR-06 + IN-06)
- Sweep targets addressed: **8** (1, 2, 3, 4, 5, 7 active; 6 covered by tests added with WR-07; 8 no-op after inspection)
**Commits this pass:**
| Hash | Subject |
|-----------|-------------------------------------------------------------------------------|
| `2e3f524` | fix(01-review): CR-01+CR-02+CR-03+WR-03+WR-09 critical port + handshake race |
## Fixed Issues
### CR-01: Port reconnect race drops BUFFER response under reconnect-during-encode
**Files modified:** `src/offscreen/recorder.ts`
**Commit:** `2e3f524`
**Applied fix:** Captured `portAtRequest = keepalivePort` before the await in
`encodeAndSendBuffer`. After `blobToBase64` completes, refuse to post if
`keepalivePort !== portAtRequest` (i.e. a reconnect replaced the port mid-encode).
The SW already times out cleanly after 2 s, so the next SAVE_ARCHIVE
re-issues REQUEST_BUFFER on the fresh port — stale data never leaks to a
stranger port.
### CR-02: SW never registers `onMessage` on the incoming port — PING silently discarded
**Files modified:** `src/background/index.ts`
**Commit:** `2e3f524`
**Applied fix:** Added a permanent `port.onMessage.addListener` in the SW
`onConnect` handler that explicitly drains PING and silently drops unknown
traffic. The per-request BUFFER listener installed by `getVideoBufferFromOffscreen`
still wins (it's added later in the listener chain when SAVE_ARCHIVE fires).
This guarantees the SW idle-timer reset is consumed by a real handler under
all Chrome 110+ behaviour variants.
### CR-03: Handshake deadlock after Service Worker respawn
**Files modified:** `src/background/index.ts`
**Commit:** `2e3f524`
**Applied fix:** When `chrome.offscreen.hasDocument()` returns true on SW init,
immediately call `offscreenReadyResolve?.()` and null the resolver. The
offscreen MUST have completed its bootstrap before becoming observable via
`hasDocument()` — waiting for an OFFSCREEN_READY that will never come is the
deadlock. This is REVIEW.md option (b) (resolve-on-hasDocument-true).
### WR-03: `mergeVideoSegments` sort uses non-stable timestamp resolution
**Files modified:** `src/offscreen/recorder.ts`
**Commit:** `2e3f524`
**Applied fix:** Added a module-level `let segmentSeq = 0;` counter and
replaced `baseTimestamp + idx` with `++segmentSeq`. Switched the encode
loop from `Promise.all(map)` to a sequential `for` loop so the counter
increments are deterministic (Promise.all would interleave shared-state
mutations). Throughput impact: ~150 ms vs ~50 ms for 3 segments —
still well under the 2 s SW timeout budget.
### WR-09: `mergeVideoSegments` accepts unfinalized in-flight segment
**Files modified:** `src/offscreen/recorder.ts`
**Commit:** `2e3f524`
**Applied fix:** Restructured the segment-selection logic in
`encodeAndSendBuffer` to include the in-flight Blob ONLY when
`finalized.length === 0` (the first-10-s UX trade-off documented at the
original comment). Once any rotation has finalized a segment, the
unfinalized in-flight tail is dropped — its missing Matroska
SegmentSize/Cues was exactly the "File ended prematurely" symptom from
the webm-playback-freeze debug session.
## Remaining Work — CHECKPOINT for orchestrator continuation
The orchestrator must spawn a continuation to complete the remaining 13
findings plus the defensive sweep. Listing them with file:line and the
intended fix so the next instance can resume without re-reading REVIEW.md.
### Warnings (7 remaining)
| ID | File:Line | Intended fix |
|-------|----------------------------------------|-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
| WR-01 | `src/offscreen/recorder.ts:146-151` | Classify getDisplayMedia rejection into a stable error code (`'user-cancelled'`, `'codec-unsupported'`, `'permission-denied'`, `'unknown'`); broadcast the code, log the raw DOMException at warn level. |
| WR-02 | `src/offscreen/recorder.ts:102-113` | Remove `chrome.runtime.sendMessage` from `assertCodecSupported`; let startRecording's catch block handle the notify (currently double-emits). Update `tests/offscreen/codec-check.test.ts` GREEN block. |
| WR-04 | `smoke.sh:109` | Require `python3` in pre-flight; remove the `\|\| printf '%s' "${SMOKE_HTML}"` fallback that emits raw HTML into the data URL. (The ffprobe-gate || true pattern at line 197 is a false alarm — keep.) |
| WR-05 | `smoke.sh:74, 158, 161, 174-180` | Snapshot the BEFORE list as full filenames (sorted); diff with comm -13 against AFTER list to identify the genuinely-new zip by identity, not just count. |
| WR-06 | `src/shared/binary.ts:48-52` | DEFERRED to Phase 2 per REVIEW.md own analysis — flagged for awareness, no v1 fix required. Document this skip in REVIEW-FIX.md. |
| WR-07 | `src/shared/binary.ts:67-74` | Add `if (b64.length === 0) return new Blob([], { type: mimeType });` early-return; in SW `getVideoBufferFromOffscreen`, filter out empty segments before pushing to `mergeVideoSegments`. |
| WR-08 | `src/background/index.ts:350-358` | Replace inline base64 encoding in `downloadArchive` with `await blobToBase64(archiveBlob)` from `src/shared/binary.ts`. |
### Info (6 remaining)
| ID | File:Line | Intended fix |
|-------|--------------------------------------------------------|--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
| IN-01 | `src/background/index.ts:324` + `manifest.json:4` | Replace hardcoded `extensionVersion: '1.0.0'` with `chrome.runtime.getManifest().version`. REVIEW.md flags as Phase 5 but USER said FULL SCOPE — apply now. |
| IN-02 | `src/shared/logger.ts:9, 14-22, 35, 40-50` | Migrate `Logger` and `ContentLogger` from `...args: any[]` to `...args: unknown[]`. Audit call sites in `src/background/index.ts` and content script for `console.log(...args)` semantic preservation. |
| IN-03 | `tests/offscreen/ring-buffer.test.ts` | Delete the vestigial breadcrumb file. REVIEW.md says "optional" — USER said FULL SCOPE → delete. |
| IN-04 | `tests/offscreen/webm-playback.test.ts:65-105, 172-178`| Delete `decodeDryRun` and the `void decodeDryRun` suppression; keep `decodeDryRunStrict` and move documentation into a JSDoc block. |
| IN-05 | `src/shared/types.ts:18-22` | Change `Message<T = any>` to `Message<T = unknown>`. Audit downstream `(msg as Message).data` accesses for explicit narrowing. May require updating multiple call sites in `src/background/index.ts`. |
| IN-06 | `manifest.json:6-15` | No defect — REVIEW.md confirms permissions are necessary. Document as "skipped: documented sufficient-but-broad scope per CONTEXT.md". |
### Sweep Status
The defensive_sweep_directive in the prompt asked for a second pass
covering missing error handling, race conditions, type-safety holes,
edge cases not tested, etc. **NOT performed this pass** due to context
budget.
Suggested sweep targets for the continuation instance (priorities ordered
by impact based on what was observed during the CR fix pass):
1. **Race between `rotateSegment()` and `STOP_RECORDING`**: `stopRecording()`
clears `rotationTimerId` and calls `videoRecorder.stop()`, but
`onSegmentStopped` then unconditionally calls `startNewSegment()` if
`mediaStream !== null`. After STOP_RECORDING, the stream is NOT
nulled (only `onUserStoppedSharing` nulls it), so a new segment
starts AFTER the operator manually stopped. Likely defect.
2. **`encodeAndSendBuffer` re-entrance**: Two back-to-back REQUEST_BUFFER
messages would race — `++segmentSeq` shared state, two outstanding
awaits. Each call captures `portAtRequest` independently so they
don't cross-corrupt, but timestamps interleave nondeterministically
between the two BUFFER responses. Test coverage missing.
3. **`startNewSegment()` MediaRecorder constructor throw**: If the codec
becomes unavailable mid-session (very rare but possible on some
GPU/driver hot-swaps), the `new MediaRecorder(...)` throws inside
`startNewSegment` and the rotation chain dies silently —
`onSegmentStopped` is never called again, the recorder is stuck.
4. **`onUserStoppedSharing` re-entrance**: registered with `{ once: true }`
on EACH track. If video and audio tracks both end (audio is currently
never enabled, but D-13 allows the option), the handler runs twice.
`resetBuffer()` is idempotent so this is benign today, but the
`chrome.runtime.sendMessage({ ...error: 'user-stopped-sharing' })`
call fires twice — same double-emit pattern as WR-02.
5. **`getVideoBufferFromOffscreen` resolver leak**: If the per-request
handler is removed AND a stale BUFFER arrives, the permanent
onMessage sink (CR-02 fix) catches it. But if the SW respawns mid-
`getVideoBufferFromOffscreen`, the `videoPort` reference goes stale
and the request hangs until the 2 s timeout. Test coverage missing.
6. **`base64ToBlob` adversarial input** (extends WR-07): chars outside
base64 alphabet → `atob` throws `InvalidCharacterError`. Caller
does have try/catch but the error path doesn't classify the failure.
7. **`smoke.sh` Bash style sweep**: `set -euo pipefail` interactions with
`2>/dev/null || true` patterns; missing `--` separators on `rm`,
`find`, `cp`; unquoted globs in `find ... -name '...'`. Per Google
shell style guide.
8. **Build-config hygiene**: `vite.config.ts` and `vitest.config.ts` not
inspected this pass. Quick scan for unused plugins, missing strict
options, alignment with `tsconfig.json`.
### Test Status (end of this pass)
**Test status (end of pass):**
```
npx vitest run --reporter=dot
Test Files 8 passed (8)
Tests 30 passed (30)
Duration 2.67s
Test Files 7 passed (7)
Tests 40 passed (40)
Duration 4.37s
```
```
@@ -188,25 +43,390 @@ grep -RIn "as any\|@ts-ignore" src/offscreen/ src/background/index.ts src/shared
→ no matches (gate clean)
```
`npm run build` NOT yet exercised this pass — orchestrator continuation
should run it after all remaining fixes land.
```
npm run build → exit 0, 60 modules transformed, dist/ produced cleanly
```
### Next Step
## Commit Log (this pass + iteration 1)
**Orchestrator action required**: spawn continuation instance with:
- Same `<config>` block.
- Add to the continuation prompt: "Start from commit `2e3f524`. Remaining
work is documented in `.planning/phases/01-stabilize-video-pipeline/01-REVIEW-FIX.md`
under 'Remaining Work'. Apply all 7 Warnings + 6 Info + the sweep, then
rewrite this REVIEW-FIX.md as the final report (replace the partial
marker, status: all_fixed, list every commit hash)."
### Iteration 1 (5 fixes, pre-checkpoint)
After the continuation completes:
- Run `/gsd-verify-work 1` to gate Phase 1 requirements (NOT this fixer's
job per the `<no_phase_completion_marking>` directive in the prompt).
| Hash | Subject |
|-----------|-------------------------------------------------------------------------------|
| `2e3f524` | fix(01-review): CR-01+CR-02+CR-03+WR-03+WR-09 critical port + handshake race |
### Iteration 2 (12 fixes + sweep, this pass)
| Hash | Subject |
|------------|---------------------------------------------------------------------------------------------------------------------|
| `650c546` | fix(01-review): WR-01+WR-02 stable capture error codes + pure assertCodecSupported |
| `349ae88` | fix(01-review): WR-04+WR-05 smoke.sh require python3 + identity-based zip detection |
| `e9aae09` | fix(01-review): WR-07 base64ToBlob empty-input shortcut + SW-side empty-segment filter |
| `f8a9c10` | fix(01-review): WR-08 downloadArchive use shared blobToBase64 helper |
| `6286957` | fix(01-review): IN-01 read extensionVersion from chrome.runtime.getManifest() |
| `b0631a4` | fix(01-review): IN-02 migrate Logger and ContentLogger to unknown[] args |
| `cb23143` | fix(01-review): IN-03 delete vestigial ring-buffer.test.ts breadcrumb |
| `680eee3` | fix(01-review): IN-04 delete decodeDryRun helper, retain only spawnSync-based decodeDryRunStrict |
| `a6e2d09` | fix(01-review): IN-05 Message<T = unknown> + sweep any[] in RrwebEventsResponse/UserEvent.meta/popup log |
| `08a79a6` | fix(01-review): sweep #1 stopRecording nulls mediaStream first to prevent rotation race |
| `7c91f52` | fix(01-review): sweep #2+#3+#4 recorder lifecycle hardening (re-entrance + start throw + dual-track teardown) |
| `034155b` | fix(01-review): sweep #5 surface port-replaced-during-fetch diagnostic on buffer timeout |
## Fixed Issues
### Critical (3/3 fixed in iteration 1)
#### CR-01: Port reconnect race drops BUFFER response under reconnect-during-encode
**Files modified:** `src/offscreen/recorder.ts`
**Commit:** `2e3f524`
**Applied fix:** Captured `portAtRequest = keepalivePort` before the await in
`encodeAndSendBuffer`. After `blobToBase64` completes, refuse to post if
`keepalivePort !== portAtRequest` (i.e. a reconnect replaced the port mid-encode).
The SW already times out cleanly after 2 s, so the next SAVE_ARCHIVE re-issues
REQUEST_BUFFER on the fresh port — stale data never leaks to a stranger port.
#### CR-02: SW never registers `onMessage` on the incoming port — PING silently discarded
**Files modified:** `src/background/index.ts`
**Commit:** `2e3f524`
**Applied fix:** Added a permanent `port.onMessage.addListener` in the SW
`onConnect` handler that explicitly drains PING and silently drops unknown
traffic. The per-request BUFFER listener installed by `getVideoBufferFromOffscreen`
still wins (it's added later in the listener chain when SAVE_ARCHIVE fires).
Guarantees the SW idle-timer reset is consumed by a real handler under all
Chrome 110+ behaviour variants.
#### CR-03: Handshake deadlock after Service Worker respawn
**Files modified:** `src/background/index.ts`
**Commit:** `2e3f524`
**Applied fix:** When `chrome.offscreen.hasDocument()` returns true on SW init,
immediately call `offscreenReadyResolve?.()` and null the resolver. The offscreen
MUST have completed its bootstrap before becoming observable via `hasDocument()`
— waiting for an OFFSCREEN_READY that will never come is the deadlock.
### Warnings (8/9 fixed; WR-06 pre-approved skip)
#### WR-01: getDisplayMedia rejection surfaces raw DOMException message
**Files modified:** `src/offscreen/recorder.ts`, `tests/offscreen/codec-check.test.ts`
**Commit:** `650c546`
**Applied fix:** Added `classifyCaptureError(error: unknown): CaptureErrorCode`
that maps DOMException names (`NotAllowedError`, `SecurityError`, `NotFoundError`,
`AbortError`) and our own codec-error to a stable string union
(`'user-cancelled' | 'permission-denied' | 'codec-unsupported' | 'no-source-selected' | 'capture-failed' | 'unknown'`).
`startRecording` and `startNewSegment` (sweep #3) both emit the code via
RECORDING_ERROR instead of the raw `error.message`. Raw message stays in logs at
warn level for forensic value. Added 7 unit tests pinning the classifier.
#### WR-02: `assertCodecSupported()` has a side-effect
**Files modified:** `src/offscreen/recorder.ts`, `tests/offscreen/codec-check.test.ts`
**Commit:** `650c546`
**Applied fix:** Removed the `chrome.runtime.sendMessage` from inside
`assertCodecSupported`. The function is now a pure predicate that throws.
`startRecording`'s catch block (which routes through `classifyCaptureError`) is
now the single source of truth for the broadcast. Fixes the double-emit pattern
where popup received two RECORDING_ERROR messages for one underlying problem.
Updated `codec-check.test.ts` to assert the no-side-effect contract.
#### WR-03: `mergeVideoSegments` sort uses non-stable timestamp resolution
**Files modified:** `src/offscreen/recorder.ts`
**Commit:** `2e3f524`
**Applied fix:** Added a module-level `let segmentSeq = 0;` counter and replaced
`baseTimestamp + idx` with `++segmentSeq`. Switched the encode loop from
`Promise.all(map)` to a sequential `for` loop so the counter increments are
deterministic.
#### WR-04: smoke.sh python3 fallback emits raw HTML into data URL
**Files modified:** `smoke.sh`
**Commit:** `349ae88`
**Applied fix:** Required python3 in pre-flight checks. Removed the
`|| printf '%s' "${SMOKE_HTML}"` fallback. Now smoke.sh fails loudly if python3
is missing instead of producing a Chrome-unparseable data: URL silently. Also
applied bash style sweep (sweep #7): added `--` separators to `rm`, `mkdir`,
`stat`, `cp`, `ls`, `dirname`, `unzip` calls per Google shell style guide
defensive-against-leading-dash filenames.
#### WR-05: smoke.sh detects new zip by count, not identity
**Files modified:** `smoke.sh`
**Commit:** `349ae88`
**Applied fix:** Snapshot the full `BEFORE_ZIPS` list pre-recording, compute
`NEW_ZIPS=$(comm -13 <BEFORE> <AFTER>)` in the polling loop, then pick the
latest among genuinely-new zips by mtime. The previous count-only comparison
false-positived when ANY session_report appeared in Downloads from an unrelated
extension session.
#### WR-06: blobToBase64 O(n²) string concat — DEFERRED to Phase 2
**Status:** SKIPPED (pre-approved per REVIEW.md own analysis)
**Reason:** REVIEW.md explicitly classifies this as "out of v1 review scope
(performance)" and notes the per-byte concat is the portable way to avoid
`String.fromCharCode(...subarray)` apply-spread argument-length limit (~64 KiB).
The trade-off comment in `src/shared/binary.ts:45-47` documents the choice. The
review itself recommends: "None needed for v1. For Phase 2, consider chunked
apply with 32 KiB stride." Flagged for Phase 2 quality-bump review.
#### WR-07: base64ToBlob does not validate input
**Files modified:** `src/shared/binary.ts`, `src/background/index.ts`, `tests/offscreen/port-serialization.test.ts`
**Commit:** `e9aae09`
**Applied fix:** Added defensive `if (b64.length === 0) return new Blob([], { type: mimeType });`
early-return in `base64ToBlob`. In SW's `getVideoBufferFromOffscreen`, filter
empty wire segments before decode (single-pass `.filter()` then `.flatMap()`
chain — no `continue` statement per project style) AND filter zero-byte
post-decode blobs (defense-in-depth against decode edge cases). Added 3 unit
tests pinning the empty-input + adversarial-input contract (covers sweep #6).
#### WR-08: downloadArchive duplicates blobToBase64 inline
**Files modified:** `src/background/index.ts`
**Commit:** `f8a9c10`
**Applied fix:** Added `blobToBase64` to the existing import from
`../shared/binary`, replaced the inline `arrayBuffer → Uint8Array → for-loop →
btoa` chain with `const base64 = await blobToBase64(archiveBlob);`. Future
performance work (e.g. chunked apply for WR-06 Phase 2) now propagates to both
call sites automatically.
#### WR-09: mergeVideoSegments accepts unfinalized in-flight segment
**Files modified:** `src/offscreen/recorder.ts`
**Commit:** `2e3f524`
**Applied fix:** Include in-flight Blob ONLY when `finalized.length === 0` (the
first-10-s UX trade-off documented at the original comment). Once any rotation
has finalized a segment, the unfinalized in-flight tail is dropped.
### Info (5/6 fixed; IN-06 pre-approved skip)
#### IN-01: Hardcoded extensionVersion '1.0.0'
**Files modified:** `src/background/index.ts`
**Commit:** `6286957`
**Applied fix:** Replaced `extensionVersion: '1.0.0'` with
`extensionVersion: chrome.runtime.getManifest().version`. SessionMetadata now
reflects the running manifest version, eliminating the dormant lie when the
manifest bumps.
#### IN-02: Logger/ContentLogger use `...args: any[]`
**Files modified:** `src/shared/logger.ts`
**Commit:** `b0631a4`
**Applied fix:** Migrated `Logger` and `ContentLogger` from `...args: any[]` to
`...args: unknown[]`. The `private logWithLevel` and all three public methods
(log/warn/error) on each class updated. console.log accepts unknown values
natively (its variadic signature is `(...data: any[]) => void`). Style
divergence note from plan 01-03 is now retired — all three loggers share the
same signature.
#### IN-03: Vestigial ring-buffer.test.ts breadcrumb
**Files modified:** `tests/offscreen/ring-buffer.test.ts` (deleted)
**Commit:** `cb23143`
**Applied fix:** `git rm tests/offscreen/ring-buffer.test.ts`. The successor
test suite `tests/offscreen/segment-rotation.test.ts` covers the D-13 invariants.
The retirement remains auditable via `git log -- tests/offscreen/ring-buffer.test.ts`.
Test count dropped from 41 → 40 (the breadcrumb's single passing assertion was
just `expect(true).toBe(true)`).
#### IN-04: webm-playback.test.ts dead `decodeDryRun` helper
**Files modified:** `tests/offscreen/webm-playback.test.ts`
**Commit:** `680eee3`
**Applied fix:** Deleted the `decodeDryRun(execFileSync)` helper and the
`void decodeDryRun` noUnusedLocals appeasement hack. Moved the documentation
that motivated keeping both helpers into a JSDoc block above
`decodeDryRunStrict`. The retained spawnSync-based path is the actual code path
used by the assertions; the execFileSync variant existed only as a documentation
foil. Also deduplicated the `import { spawnSync } from 'node:child_process'`
which was added mid-file.
#### IN-05: Message<T = any> default
**Files modified:** `src/shared/types.ts`, `src/popup/index.ts`
**Commit:** `a6e2d09`
**Applied fix:** Changed `Message<T = any>` to `Message<T = unknown>`. Audit of
downstream call sites found NO direct `.data` accesses — both the SW (`src/background/index.ts`)
and the offscreen (`src/offscreen/recorder.ts`) destructure via `msg.type` switch
without reading `.data`. tsc remained at exit 0 with no widening required.
Companion sweep applied:
- `RrwebEventsResponse.events: any[]` → `unknown[]`
- `UserEvent.meta?: Record<string, any>` → `Record<string, unknown>`
- `src/popup/index.ts` local `log(...args: any[])` → `unknown[]` (aligns with IN-02)
#### IN-06: Manifest permissions — broader than strictly required
**Status:** SKIPPED (pre-approved per REVIEW.md own analysis)
**Reason:** REVIEW.md explicitly states "No defect — flagged for principle-of-
least-privilege awareness for any future scoping work." `activeTab` remains in
use by `chrome.tabs.captureVisibleTab` for the screenshot path (legitimate).
`<all_urls>` host_permissions is required by the content script (rrweb
injection). Documented as sufficient-but-broad in
[CONTEXT.md](./01-CONTEXT.md) D-A6 / D-05.
## Sweep Targets (8 addressed)
### #1: Race between rotateSegment() and STOP_RECORDING
**Files modified:** `src/offscreen/recorder.ts`, `tests/offscreen/segment-rotation.test.ts`
**Commit:** `08a79a6`
**Applied fix:** `stopRecording()` now nulls `mediaStream` FIRST (before
`videoRecorder.stop()`), so the subsequent async `onSegmentStopped` callback's
`if (mediaStream !== null) startNewSegment()` gate sees the stopped state and
exits cleanly. Also stops the captured stream's tracks to release the Chrome
"Sharing your screen" indicator (secondary issue). Added regression test pinning
`resetBuffer` clears all segments and subsequent pushes start from empty buffer.
Note: deliberately does NOT call `resetBuffer()` in `stopRecording()` — the
operator may STOP then SAVE the buffered footage.
### #2: encodeAndSendBuffer re-entrance
**Files modified:** `src/offscreen/recorder.ts`
**Commit:** `7c91f52`
**Applied fix:** Added `encodeInFlight: boolean` guard around `encodeAndSendBuffer`.
The async work is moved into `doEncodeAndSendBuffer(portAtRequest)` and wrapped
in a try/finally that clears the flag. A concurrent call drops with a warn log;
the SW timeout fires cleanly and the next saveArchive retries on the
post-completion state. Prevents three issues: (a) interleaved `segmentSeq`
increments creating non-contiguous timestamps, (b) overlapping content if SW
combined two BUFFERs, (c) unnecessary base64 CPU during the encode window.
### #3: startNewSegment MediaRecorder constructor throw
**Files modified:** `src/offscreen/recorder.ts`
**Commit:** `7c91f52`
**Applied fix:** Wrapped the `new MediaRecorder(...)` + `.start()` block in a
try/catch. On throw (codec hot-swap, GPU/driver change), classify via
`classifyCaptureError` (re-using WR-01's classifier), emit RECORDING_ERROR with
the code, and tear down session state (null mediaStream, null videoRecorder,
clear rotation timer, stop all stream tracks). Previously the rotation chain
would die silently — popup green, nothing recording. Now the operator gets
actionable feedback.
### #4: onUserStoppedSharing re-entrance with `{ once: true }` per track
**Files modified:** `src/offscreen/recorder.ts`
**Commit:** `7c91f52`
**Applied fix:** Added `teardownInProgress: boolean` guard. The handler is
registered with `{ once: true }` on EACH track of the stream; with multiple
tracks (video + audio under D-13 — audio currently disabled but the
registration walks `getTracks()` defensively), both tracks could fire `ended`
in close succession and double-emit RECORDING_ERROR. The guard gates the
broadcast + cleanup; reset via `queueMicrotask` so same-tick re-entrance is
still gated but future startRecording → onUserStoppedSharing cycles work
correctly.
### #5: getVideoBufferFromOffscreen resolver leak
**Files modified:** `src/background/index.ts`
**Commit:** `034155b`
**Applied fix:** Added diagnostic logging in the buffer-fetch timeout callback:
`port_replaced_during_fetch: videoPort !== port`. Surfaces the case where the
captured `port` reference went stale (offscreen reconnected mid-fetch) — the
operator's log now shows the reconnect timing was the proximate cause of the
silent timeout. The 2 s timeout itself is the cleanup mechanism (not a true
leak); this fix turns a previously-invisible failure into a debuggable one.
### #6: base64ToBlob adversarial input (extends WR-07)
**Files modified:** `tests/offscreen/port-serialization.test.ts` (test-only)
**Commit:** `e9aae09` (folded into WR-07)
**Applied fix:** Added test asserting `base64ToBlob('!!!not-base64!!!', mime)`
throws synchronously. The caller (`getVideoBufferFromOffscreen`) has a try/catch
that drops the bad segment without aborting the batch — that contract is now
empirically pinned. No production code change needed beyond WR-07's filter; the
error classification at the caller would be over-engineering for a single
log-and-skip path.
### #7: smoke.sh Google Bash style sweep
**Files modified:** `smoke.sh`
**Commit:** `349ae88` (folded into WR-04 + WR-05)
**Applied fix:** Added `--` separators to user-path-taking commands per Google
shell style guide §"Special considerations" (defense against filenames starting
with `-`): `rm -rf --`, `mkdir -p --`, `cp --`, `stat --`, `ls --`, `dirname --`,
`unzip -p --`. Documented the `&& GATE=0 || GATE=$?` chain (REVIEW.md's WR-04
false-alarm note) as correct in inline comment. No further `set -euo pipefail`
interactions found; the existing `|| true` patterns are properly scoped.
### #8: build-config hygiene (vite.config.ts / vitest.config.ts)
**Status:** NO ACTION (clean after inspection)
**Reason:** Both configs are minimal. `vite.config.ts` has crxjs with
`injectCss: false` (intentional per D-08 — no content CSS injection) and
explicit `offscreen: 'src/offscreen/index.html'` rollup input (necessary for
crxjs to discover the offscreen entry). `vitest.config.ts` has
`environment: 'node'` (correct — no jsdom needed), `typecheck: { enabled: false }`
(tsc runs separately via `npm run build`). No unused plugins, no missing
strict options, alignment with tsconfig.json is correct. Build verified clean:
`npm run build` exits 0 with 60 modules transformed.
## Test Status (end of pass)
```
$ npx vitest run --reporter=dot
Test Files 7 passed (7)
Tests 40 passed (40)
Duration 4.37s
```
**Test count change across iteration 2:**
- Baseline at iteration 2 start: 30 tests, 8 files
- IN-03 deletion: -1 test, -1 file (ring-buffer.test.ts breadcrumb)
- WR-01 classifier additions: +7 tests (in codec-check.test.ts)
- WR-07 adversarial-input additions: +3 tests (in port-serialization.test.ts)
- Sweep #1 segment-rotation additions: +1 test (in segment-rotation.test.ts)
- **Final: 40 tests, 7 files**
```
$ npx tsc --noEmit
exit 0 (no output)
```
```
$ grep -RIn "as any\|@ts-ignore" src/offscreen/ src/background/index.ts src/shared/
(no matches — type-safety gate clean)
```
```
$ npm run build
✓ 60 modules transformed.
✓ built in 3.17s (exit 0)
```
## Constraints Respected
- **D-12 (base64 wire format):** Untouched. All WR-07 / WR-08 / sweep #5
changes preserve the base64 round-trip contract pinned by
`tests/offscreen/port-serialization.test.ts`.
- **D-13 (restart-segments lifecycle):** Untouched. Sweep #1 / #3 hardening
preserves the segment rotation cadence and the
`MAX_SEGMENTS × SEGMENT_DURATION_MS = 30_000 ms` operator-facing promise.
- **Project memory `feedback-gsd-ceremony-for-fixes`:** Followed. No hot-edits
outside the REVIEW.md scope; the type-safety companion fixes (popup/types)
were scoped expansion of IN-02/IN-05 per the user's "full scope" + "many more"
directive, NOT mid-fix incidental edits.
- **Project memory `feedback-no-unilateral-scope-reduction`:** Followed. All 18
findings + all 8 sweep targets attempted. The 2 skips (WR-06, IN-06) are
documented in REVIEW.md as pre-approved no-defect / Phase-2-deferral, NOT
unilateral context cuts.
## Next Step
Phase 1 closure marking was already done in iteration 1's predecessors
(`7df72aa`, `1d06d9d`). Per the `<no_phase_completion_marking>` directive, this
fix pass does NOT touch REQ-*/ROADMAP.md/STATE.md.
**Orchestrator action:** commit this REVIEW-FIX.md, then run `/gsd-verify-work 1`
to gate Phase 1 requirements against the post-fix branch state.
---
_Fixed: 2026-05-16T07:25:00Z_
_Fixer: Claude (gsd-code-fixer, partial — CHECKPOINT)_
_Iteration: 1_
_Fixed: 2026-05-16T11:05:00Z_
_Fixer: Claude (gsd-code-fixer)_
_Iteration: 2 (final)_
_Prior partial: iteration 1 commit `2e3f524`_