test(desktop): sample remount restore on painted frames

[1625567290]

Summary

The "returning to a live conversation settles output accumulated while away" e2e sampled .maka-bubble-streaming on every document.body mutation. That observer sees React commit intermediates that never paint, so the "no partial catch-up" check depended on how much unrelated shell churn happened during remount. It passed in isolation and failed when the rest of the file had already warmed the machine.

The restore is now sampled on animation frames of the streaming bubble only. A real multi-frame replay would still fail; an intra-frame commit that the user never sees would not. The assertion also requires at least one painted sample to contain the settled background text, so a silent no-op observer cannot pass.

Fixes#3061

Verification

• npx biome check apps/desktop/e2e/streaming-remount.spec.ts — clean
• Isolated: playwright test e2e/streaming-remount.spec.ts --grep "accumulated while away" — pass
• Whole file after the change: 3/3, 3/3, plus one earlier whole-file run where this case still passed and an unrelated sibling (keeps a completed reply after an interrupted turn) timed out waiting for 停止 (pre-existing load flake, not this observer)
• Before the change, this machine did not reproduce the original body-observer failure in 5 whole-file runs; the fix follows the diagnosis in the issue (observation too coarse), not a local red reproduction of that exact assertion

Not run: full desktop e2e suite, npm --workspace @maka/desktop test (no production code change)

AI use

Select exactly one:

• No generative tool made a substantive contribution
• Generative tooling made a substantive contribution

Tool(s) and scope:

Grok authored the test change and this PR description.

Checklist

• Tests cover the change and fail without it
• Lint, format, typecheck and the affected suites pass locally

Does this PR entail a change in behavior?

• Yes — described under Summary above
• No

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: 7cd6910e-da76-4235-9bf8-2171be824736

📥 Commits

Reviewing files that changed from the base of the PR and between 1339419 and 81add45.

📒 Files selected for processing (1)

• apps/desktop/e2e/streaming-remount.spec.ts

🚧 Files skipped from review as they are similar to previous changes (1)

• apps/desktop/e2e/streaming-remount.spec.ts

Included review availability: Your plan includes up to 3 reviews per rolling hour; 1 remains after this review.

---

📝 Walkthrough

Summary

• Fixes intermittent failures in the desktop background-restore E2E test.
• Samples .maka-bubble-streaming on animation frames instead of all document.body mutations.
• Avoids recording React commit intermediates that never paint.
• Confirms that a painted sample contains the settled background text.
• Confirms that no painted sample contains partially accumulated background output.

Design assessment

The change extends the existing E2E test. It does not create a parallel production path or change public contracts.

The frame sampler, sample deduplication, animation tracking, stoppable lifecycle, and next-frame stop are necessary to observe settled painted output reliably. No safe deletion or simplification is evident without weakening regression coverage.

Validation

• Biome checks were run.
• Repeated isolated and whole-file Playwright runs were performed.
• The full desktop E2E suite and desktop unit tests were not run because production code did not change.
• Required-check status remains unverified because direct current diff and status evidence is unavailable.

Review-relevant risks

No protected-area effect was identified in the current diff.

Walkthrough

The background-restore end-to-end test replaces document-wide mutation observation with a stoppable requestAnimationFrame sampler. It records distinct streaming text, tracks active animations, and stops sampling after the next animation frame.

Changes

Streaming restore test

Layer / File(s)	Summary
Frame-based output sampling apps/desktop/e2e/streaming-remount.spec.ts	The test samples the streaming bubble on animation frames, deduplicates text, tracks unfinished animations, waits for one final frame, stops the sampler, and validates restored output.

Estimated code review effort: 2 (Simple) | ~10 minutes

Merge Risk:⚪ Minimal · up to 81add

This localized test-only change adjusts how remount output is sampled without changing product behavior, and no actionable merge-blocking risk remains after normal checks and review.

Possibly related PRs

• maka-agent/maka-agent#3057: Modifies the same streaming-remount end-to-end test and related output-restoration assertions.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly describes the main change: sampling desktop remount restoration on painted animation frames.
Description check	✅ Passed	The description follows the template, explains the problem and solution, lists verification, documents AI use, and records checklist decisions.
Linked Issues check	✅ Passed	The test changes address issue #3061 by sampling painted frames and requiring settled output without partial catch-up states.
Out of Scope Changes check	✅ Passed	The changes are limited to the related desktop streaming-remount E2E test and its observation logic.
Ai Use Disclosure	✅ Passed	The PR selects generative tooling and names Grok with scope. Both introduced commits contain standalone Generated-by: Grok trailers.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

_{Comment @coderabbitai help to get the list of available commands.}

[coderabbitai]

Actionable comments posted: 1

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: cc9670e7-a884-4981-8b6d-710594b20c29

📥 Commits

Reviewing files that changed from the base of the PR and between 9fbdc99 and 1339419.

📒 Files selected for processing (1)

• apps/desktop/e2e/streaming-remount.spec.ts

Included review availability: Your plan includes up to 3 reviews per rolling hour; 2 remain after this review.

[Astro-Han]

Fast-path merge

This change qualifies for the self-merge fast path: it is low impact and easy to reverse (test-only change sampling the remount restore on painted animation frames instead of every DOM mutation), does not touch the protected areas (Runtime Host execution authority, @maka/eval semantics, public Eval CLI, security, licensing, releases, governance), and passes all required checks.

The latest head was independently reviewed by a read-only subagent (ollama-cloud/deepseek-v4-flash:high) with no open P0/P1/P2 findings. Human contributor @yuhan reviewed the final diff and chose the fast path.

[1625567290]

Summary

The "returning to a live conversation settles output accumulated while away" e2e sampled .maka-bubble-streaming on every document.body mutation. That observer sees React commit intermediates that never paint, so the "no partial catch-up" check depended on how much unrelated shell churn happened during remount. It passed in isolation and failed when the rest of the file had already warmed the machine.

The restore is now sampled on animation frames of the streaming bubble only. A real multi-frame replay would still fail; an intra-frame commit that the user never sees would not. The assertion also requires at least one painted sample to contain the settled background text, so a silent no-op observer cannot pass.

Fixes#3061

Verification

• npx biome check apps/desktop/e2e/streaming-remount.spec.ts — clean
• Isolated: playwright test e2e/streaming-remount.spec.ts --grep "accumulated while away" — pass
• Whole file after the change: 3/3, 3/3, plus one earlier whole-file run where this case still passed and an unrelated sibling (keeps a completed reply after an interrupted turn) timed out waiting for 停止 (pre-existing load flake, not this observer)
• Before the change, this machine did not reproduce the original body-observer failure in 5 whole-file runs; the fix follows the diagnosis in the issue (observation too coarse), not a local red reproduction of that exact assertion

Not run: full desktop e2e suite, npm --workspace @maka/desktop test (no production code change)

AI use

Select exactly one:

• No generative tool made a substantive contribution
• Generative tooling made a substantive contribution

Tool(s) and scope:

Grok authored the test change and this PR description.

Checklist

• Tests cover the change and fail without it
• Lint, format, typecheck and the affected suites pass locally

Does this PR entail a change in behavior?

• Yes — described under Summary above
• No

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: 7cd6910e-da76-4235-9bf8-2171be824736

📥 Commits

Reviewing files that changed from the base of the PR and between 1339419 and 81add45.

📒 Files selected for processing (1)

• apps/desktop/e2e/streaming-remount.spec.ts

🚧 Files skipped from review as they are similar to previous changes (1)

• apps/desktop/e2e/streaming-remount.spec.ts

Included review availability: Your plan includes up to 3 reviews per rolling hour; 1 remains after this review.

---

📝 Walkthrough

Summary

• Fixes intermittent failures in the desktop background-restore E2E test.
• Samples .maka-bubble-streaming on animation frames instead of all document.body mutations.
• Avoids recording React commit intermediates that never paint.
• Confirms that a painted sample contains the settled background text.
• Confirms that no painted sample contains partially accumulated background output.

Design assessment

The change extends the existing E2E test. It does not create a parallel production path or change public contracts.

The frame sampler, sample deduplication, animation tracking, stoppable lifecycle, and next-frame stop are necessary to observe settled painted output reliably. No safe deletion or simplification is evident without weakening regression coverage.

Validation

• Biome checks were run.
• Repeated isolated and whole-file Playwright runs were performed.
• The full desktop E2E suite and desktop unit tests were not run because production code did not change.
• Required-check status remains unverified because direct current diff and status evidence is unavailable.

Review-relevant risks

No protected-area effect was identified in the current diff.

Walkthrough

The background-restore end-to-end test replaces document-wide mutation observation with a stoppable requestAnimationFrame sampler. It records distinct streaming text, tracks active animations, and stops sampling after the next animation frame.

Changes

Streaming restore test

Layer / File(s)	Summary
Frame-based output sampling apps/desktop/e2e/streaming-remount.spec.ts	The test samples the streaming bubble on animation frames, deduplicates text, tracks unfinished animations, waits for one final frame, stops the sampler, and validates restored output.

Estimated code review effort: 2 (Simple) | ~10 minutes

Merge Risk:⚪ Minimal · up to 81add

This localized test-only change adjusts how remount output is sampled without changing product behavior, and no actionable merge-blocking risk remains after normal checks and review.

Possibly related PRs

• maka-agent/maka-agent#3057: Modifies the same streaming-remount end-to-end test and related output-restoration assertions.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly describes the main change: sampling desktop remount restoration on painted animation frames.
Description check	✅ Passed	The description follows the template, explains the problem and solution, lists verification, documents AI use, and records checklist decisions.
Linked Issues check	✅ Passed	The test changes address issue #3061 by sampling painted frames and requiring settled output without partial catch-up states.
Out of Scope Changes check	✅ Passed	The changes are limited to the related desktop streaming-remount E2E test and its observation logic.
Ai Use Disclosure	✅ Passed	The PR selects generative tooling and names Grok with scope. Both introduced commits contain standalone Generated-by: Grok trailers.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

_{Comment @coderabbitai help to get the list of available commands.}

[coderabbitai]

Actionable comments posted: 1

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: cc9670e7-a884-4981-8b6d-710594b20c29

📥 Commits

Reviewing files that changed from the base of the PR and between 9fbdc99 and 1339419.

📒 Files selected for processing (1)

• apps/desktop/e2e/streaming-remount.spec.ts

Included review availability: Your plan includes up to 3 reviews per rolling hour; 2 remain after this review.

[Astro-Han]

Fast-path merge

This change qualifies for the self-merge fast path: it is low impact and easy to reverse (test-only change sampling the remount restore on painted animation frames instead of every DOM mutation), does not touch the protected areas (Runtime Host execution authority, @maka/eval semantics, public Eval CLI, security, licensing, releases, governance), and passes all required checks.

The latest head was independently reviewed by a read-only subagent (ollama-cloud/deepseek-v4-flash:high) with no open P0/P1/P2 findings. Human contributor @yuhan reviewed the final diff and chose the fast path.

[1625567290]

Summary

The "returning to a live conversation settles output accumulated while away" e2e sampled .maka-bubble-streaming on every document.body mutation. That observer sees React commit intermediates that never paint, so the "no partial catch-up" check depended on how much unrelated shell churn happened during remount. It passed in isolation and failed when the rest of the file had already warmed the machine.

The restore is now sampled on animation frames of the streaming bubble only. A real multi-frame replay would still fail; an intra-frame commit that the user never sees would not. The assertion also requires at least one painted sample to contain the settled background text, so a silent no-op observer cannot pass.

Fixes#3061

Verification

• npx biome check apps/desktop/e2e/streaming-remount.spec.ts — clean
• Isolated: playwright test e2e/streaming-remount.spec.ts --grep "accumulated while away" — pass
• Whole file after the change: 3/3, 3/3, plus one earlier whole-file run where this case still passed and an unrelated sibling (keeps a completed reply after an interrupted turn) timed out waiting for 停止 (pre-existing load flake, not this observer)
• Before the change, this machine did not reproduce the original body-observer failure in 5 whole-file runs; the fix follows the diagnosis in the issue (observation too coarse), not a local red reproduction of that exact assertion

Not run: full desktop e2e suite, npm --workspace @maka/desktop test (no production code change)

AI use

Select exactly one:

• No generative tool made a substantive contribution
• Generative tooling made a substantive contribution

Tool(s) and scope:

Grok authored the test change and this PR description.

Checklist

• Tests cover the change and fail without it
• Lint, format, typecheck and the affected suites pass locally

Does this PR entail a change in behavior?

• Yes — described under Summary above
• No

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: 7cd6910e-da76-4235-9bf8-2171be824736

📥 Commits

Reviewing files that changed from the base of the PR and between 1339419 and 81add45.

📒 Files selected for processing (1)

• apps/desktop/e2e/streaming-remount.spec.ts

🚧 Files skipped from review as they are similar to previous changes (1)

• apps/desktop/e2e/streaming-remount.spec.ts

Included review availability: Your plan includes up to 3 reviews per rolling hour; 1 remains after this review.

---

📝 Walkthrough

Summary

• Fixes intermittent failures in the desktop background-restore E2E test.
• Samples .maka-bubble-streaming on animation frames instead of all document.body mutations.
• Avoids recording React commit intermediates that never paint.
• Confirms that a painted sample contains the settled background text.
• Confirms that no painted sample contains partially accumulated background output.

Design assessment

The change extends the existing E2E test. It does not create a parallel production path or change public contracts.

The frame sampler, sample deduplication, animation tracking, stoppable lifecycle, and next-frame stop are necessary to observe settled painted output reliably. No safe deletion or simplification is evident without weakening regression coverage.

Validation

• Biome checks were run.
• Repeated isolated and whole-file Playwright runs were performed.
• The full desktop E2E suite and desktop unit tests were not run because production code did not change.
• Required-check status remains unverified because direct current diff and status evidence is unavailable.

Review-relevant risks

No protected-area effect was identified in the current diff.

Walkthrough

The background-restore end-to-end test replaces document-wide mutation observation with a stoppable requestAnimationFrame sampler. It records distinct streaming text, tracks active animations, and stops sampling after the next animation frame.

Changes

Streaming restore test

Layer / File(s)	Summary
Frame-based output sampling apps/desktop/e2e/streaming-remount.spec.ts	The test samples the streaming bubble on animation frames, deduplicates text, tracks unfinished animations, waits for one final frame, stops the sampler, and validates restored output.

Estimated code review effort: 2 (Simple) | ~10 minutes

Merge Risk:⚪ Minimal · up to 81add

This localized test-only change adjusts how remount output is sampled without changing product behavior, and no actionable merge-blocking risk remains after normal checks and review.

Possibly related PRs

• maka-agent/maka-agent#3057: Modifies the same streaming-remount end-to-end test and related output-restoration assertions.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly describes the main change: sampling desktop remount restoration on painted animation frames.
Description check	✅ Passed	The description follows the template, explains the problem and solution, lists verification, documents AI use, and records checklist decisions.
Linked Issues check	✅ Passed	The test changes address issue #3061 by sampling painted frames and requiring settled output without partial catch-up states.
Out of Scope Changes check	✅ Passed	The changes are limited to the related desktop streaming-remount E2E test and its observation logic.
Ai Use Disclosure	✅ Passed	The PR selects generative tooling and names Grok with scope. Both introduced commits contain standalone Generated-by: Grok trailers.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

_{Comment @coderabbitai help to get the list of available commands.}

[coderabbitai]

Actionable comments posted: 1

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: cc9670e7-a884-4981-8b6d-710594b20c29

📥 Commits

Reviewing files that changed from the base of the PR and between 9fbdc99 and 1339419.

📒 Files selected for processing (1)

• apps/desktop/e2e/streaming-remount.spec.ts

Included review availability: Your plan includes up to 3 reviews per rolling hour; 2 remain after this review.

[Astro-Han]

Fast-path merge

This change qualifies for the self-merge fast path: it is low impact and easy to reverse (test-only change sampling the remount restore on painted animation frames instead of every DOM mutation), does not touch the protected areas (Runtime Host execution authority, @maka/eval semantics, public Eval CLI, security, licensing, releases, governance), and passes all required checks.

The latest head was independently reviewed by a read-only subagent (ollama-cloud/deepseek-v4-flash:high) with no open P0/P1/P2 findings. Human contributor @yuhan reviewed the final diff and chose the fast path.

[1625567290]

Summary

The "returning to a live conversation settles output accumulated while away" e2e sampled .maka-bubble-streaming on every document.body mutation. That observer sees React commit intermediates that never paint, so the "no partial catch-up" check depended on how much unrelated shell churn happened during remount. It passed in isolation and failed when the rest of the file had already warmed the machine.

The restore is now sampled on animation frames of the streaming bubble only. A real multi-frame replay would still fail; an intra-frame commit that the user never sees would not. The assertion also requires at least one painted sample to contain the settled background text, so a silent no-op observer cannot pass.

Fixes#3061

Verification

• npx biome check apps/desktop/e2e/streaming-remount.spec.ts — clean
• Isolated: playwright test e2e/streaming-remount.spec.ts --grep "accumulated while away" — pass
• Whole file after the change: 3/3, 3/3, plus one earlier whole-file run where this case still passed and an unrelated sibling (keeps a completed reply after an interrupted turn) timed out waiting for 停止 (pre-existing load flake, not this observer)
• Before the change, this machine did not reproduce the original body-observer failure in 5 whole-file runs; the fix follows the diagnosis in the issue (observation too coarse), not a local red reproduction of that exact assertion

Not run: full desktop e2e suite, npm --workspace @maka/desktop test (no production code change)

AI use

Select exactly one:

• No generative tool made a substantive contribution
• Generative tooling made a substantive contribution

Tool(s) and scope:

Grok authored the test change and this PR description.

Checklist

• Tests cover the change and fail without it
• Lint, format, typecheck and the affected suites pass locally

Does this PR entail a change in behavior?

• Yes — described under Summary above
• No

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: 7cd6910e-da76-4235-9bf8-2171be824736

📥 Commits

Reviewing files that changed from the base of the PR and between 1339419 and 81add45.

📒 Files selected for processing (1)

• apps/desktop/e2e/streaming-remount.spec.ts

🚧 Files skipped from review as they are similar to previous changes (1)

• apps/desktop/e2e/streaming-remount.spec.ts

Included review availability: Your plan includes up to 3 reviews per rolling hour; 1 remains after this review.

---

📝 Walkthrough

Summary

• Fixes intermittent failures in the desktop background-restore E2E test.
• Samples .maka-bubble-streaming on animation frames instead of all document.body mutations.
• Avoids recording React commit intermediates that never paint.
• Confirms that a painted sample contains the settled background text.
• Confirms that no painted sample contains partially accumulated background output.

Design assessment

The change extends the existing E2E test. It does not create a parallel production path or change public contracts.

The frame sampler, sample deduplication, animation tracking, stoppable lifecycle, and next-frame stop are necessary to observe settled painted output reliably. No safe deletion or simplification is evident without weakening regression coverage.

Validation

• Biome checks were run.
• Repeated isolated and whole-file Playwright runs were performed.
• The full desktop E2E suite and desktop unit tests were not run because production code did not change.
• Required-check status remains unverified because direct current diff and status evidence is unavailable.

Review-relevant risks

No protected-area effect was identified in the current diff.

Walkthrough

The background-restore end-to-end test replaces document-wide mutation observation with a stoppable requestAnimationFrame sampler. It records distinct streaming text, tracks active animations, and stops sampling after the next animation frame.

Changes

Streaming restore test

Layer / File(s)	Summary
Frame-based output sampling apps/desktop/e2e/streaming-remount.spec.ts	The test samples the streaming bubble on animation frames, deduplicates text, tracks unfinished animations, waits for one final frame, stops the sampler, and validates restored output.

Estimated code review effort: 2 (Simple) | ~10 minutes

Merge Risk:⚪ Minimal · up to 81add

This localized test-only change adjusts how remount output is sampled without changing product behavior, and no actionable merge-blocking risk remains after normal checks and review.

Possibly related PRs

• maka-agent/maka-agent#3057: Modifies the same streaming-remount end-to-end test and related output-restoration assertions.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly describes the main change: sampling desktop remount restoration on painted animation frames.
Description check	✅ Passed	The description follows the template, explains the problem and solution, lists verification, documents AI use, and records checklist decisions.
Linked Issues check	✅ Passed	The test changes address issue #3061 by sampling painted frames and requiring settled output without partial catch-up states.
Out of Scope Changes check	✅ Passed	The changes are limited to the related desktop streaming-remount E2E test and its observation logic.
Ai Use Disclosure	✅ Passed	The PR selects generative tooling and names Grok with scope. Both introduced commits contain standalone Generated-by: Grok trailers.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

_{Comment @coderabbitai help to get the list of available commands.}

[coderabbitai]

Actionable comments posted: 1

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: cc9670e7-a884-4981-8b6d-710594b20c29

📥 Commits

Reviewing files that changed from the base of the PR and between 9fbdc99 and 1339419.

📒 Files selected for processing (1)

• apps/desktop/e2e/streaming-remount.spec.ts

Included review availability: Your plan includes up to 3 reviews per rolling hour; 2 remain after this review.

[Astro-Han]

Fast-path merge

This change qualifies for the self-merge fast path: it is low impact and easy to reverse (test-only change sampling the remount restore on painted animation frames instead of every DOM mutation), does not touch the protected areas (Runtime Host execution authority, @maka/eval semantics, public Eval CLI, security, licensing, releases, governance), and passes all required checks.

The latest head was independently reviewed by a read-only subagent (ollama-cloud/deepseek-v4-flash:high) with no open P0/P1/P2 findings. Human contributor @yuhan reviewed the final diff and chose the fast path.

[1625567290]

Summary

The "returning to a live conversation settles output accumulated while away" e2e sampled .maka-bubble-streaming on every document.body mutation. That observer sees React commit intermediates that never paint, so the "no partial catch-up" check depended on how much unrelated shell churn happened during remount. It passed in isolation and failed when the rest of the file had already warmed the machine.

The restore is now sampled on animation frames of the streaming bubble only. A real multi-frame replay would still fail; an intra-frame commit that the user never sees would not. The assertion also requires at least one painted sample to contain the settled background text, so a silent no-op observer cannot pass.

Fixes#3061

Verification

• npx biome check apps/desktop/e2e/streaming-remount.spec.ts — clean
• Isolated: playwright test e2e/streaming-remount.spec.ts --grep "accumulated while away" — pass
• Whole file after the change: 3/3, 3/3, plus one earlier whole-file run where this case still passed and an unrelated sibling (keeps a completed reply after an interrupted turn) timed out waiting for 停止 (pre-existing load flake, not this observer)
• Before the change, this machine did not reproduce the original body-observer failure in 5 whole-file runs; the fix follows the diagnosis in the issue (observation too coarse), not a local red reproduction of that exact assertion

Not run: full desktop e2e suite, npm --workspace @maka/desktop test (no production code change)

AI use

Select exactly one:

• No generative tool made a substantive contribution
• Generative tooling made a substantive contribution

Tool(s) and scope:

Grok authored the test change and this PR description.

Checklist

• Tests cover the change and fail without it
• Lint, format, typecheck and the affected suites pass locally

Does this PR entail a change in behavior?

• Yes — described under Summary above
• No

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: 7cd6910e-da76-4235-9bf8-2171be824736

📥 Commits

Reviewing files that changed from the base of the PR and between 1339419 and 81add45.

📒 Files selected for processing (1)

• apps/desktop/e2e/streaming-remount.spec.ts

🚧 Files skipped from review as they are similar to previous changes (1)

• apps/desktop/e2e/streaming-remount.spec.ts

Included review availability: Your plan includes up to 3 reviews per rolling hour; 1 remains after this review.

---

📝 Walkthrough

Summary

• Fixes intermittent failures in the desktop background-restore E2E test.
• Samples .maka-bubble-streaming on animation frames instead of all document.body mutations.
• Avoids recording React commit intermediates that never paint.
• Confirms that a painted sample contains the settled background text.
• Confirms that no painted sample contains partially accumulated background output.

Design assessment

The change extends the existing E2E test. It does not create a parallel production path or change public contracts.

The frame sampler, sample deduplication, animation tracking, stoppable lifecycle, and next-frame stop are necessary to observe settled painted output reliably. No safe deletion or simplification is evident without weakening regression coverage.

Validation

• Biome checks were run.
• Repeated isolated and whole-file Playwright runs were performed.
• The full desktop E2E suite and desktop unit tests were not run because production code did not change.
• Required-check status remains unverified because direct current diff and status evidence is unavailable.

Review-relevant risks

No protected-area effect was identified in the current diff.

Walkthrough

The background-restore end-to-end test replaces document-wide mutation observation with a stoppable requestAnimationFrame sampler. It records distinct streaming text, tracks active animations, and stops sampling after the next animation frame.

Changes

Streaming restore test

Layer / File(s)	Summary
Frame-based output sampling apps/desktop/e2e/streaming-remount.spec.ts	The test samples the streaming bubble on animation frames, deduplicates text, tracks unfinished animations, waits for one final frame, stops the sampler, and validates restored output.

Estimated code review effort: 2 (Simple) | ~10 minutes

Merge Risk:⚪ Minimal · up to 81add

This localized test-only change adjusts how remount output is sampled without changing product behavior, and no actionable merge-blocking risk remains after normal checks and review.

Possibly related PRs

• maka-agent/maka-agent#3057: Modifies the same streaming-remount end-to-end test and related output-restoration assertions.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly describes the main change: sampling desktop remount restoration on painted animation frames.
Description check	✅ Passed	The description follows the template, explains the problem and solution, lists verification, documents AI use, and records checklist decisions.
Linked Issues check	✅ Passed	The test changes address issue #3061 by sampling painted frames and requiring settled output without partial catch-up states.
Out of Scope Changes check	✅ Passed	The changes are limited to the related desktop streaming-remount E2E test and its observation logic.
Ai Use Disclosure	✅ Passed	The PR selects generative tooling and names Grok with scope. Both introduced commits contain standalone Generated-by: Grok trailers.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

_{Comment @coderabbitai help to get the list of available commands.}

[coderabbitai]

Actionable comments posted: 1

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: cc9670e7-a884-4981-8b6d-710594b20c29

📥 Commits

Reviewing files that changed from the base of the PR and between 9fbdc99 and 1339419.

📒 Files selected for processing (1)

• apps/desktop/e2e/streaming-remount.spec.ts

Included review availability: Your plan includes up to 3 reviews per rolling hour; 2 remain after this review.

[Astro-Han]

Fast-path merge

This change qualifies for the self-merge fast path: it is low impact and easy to reverse (test-only change sampling the remount restore on painted animation frames instead of every DOM mutation), does not touch the protected areas (Runtime Host execution authority, @maka/eval semantics, public Eval CLI, security, licensing, releases, governance), and passes all required checks.

The latest head was independently reviewed by a read-only subagent (ollama-cloud/deepseek-v4-flash:high) with no open P0/P1/P2 findings. Human contributor @yuhan reviewed the final diff and chose the fast path.

[1625567290]

Summary

The "returning to a live conversation settles output accumulated while away" e2e sampled .maka-bubble-streaming on every document.body mutation. That observer sees React commit intermediates that never paint, so the "no partial catch-up" check depended on how much unrelated shell churn happened during remount. It passed in isolation and failed when the rest of the file had already warmed the machine.

The restore is now sampled on animation frames of the streaming bubble only. A real multi-frame replay would still fail; an intra-frame commit that the user never sees would not. The assertion also requires at least one painted sample to contain the settled background text, so a silent no-op observer cannot pass.

Fixes#3061

Verification

• npx biome check apps/desktop/e2e/streaming-remount.spec.ts — clean
• Isolated: playwright test e2e/streaming-remount.spec.ts --grep "accumulated while away" — pass
• Whole file after the change: 3/3, 3/3, plus one earlier whole-file run where this case still passed and an unrelated sibling (keeps a completed reply after an interrupted turn) timed out waiting for 停止 (pre-existing load flake, not this observer)
• Before the change, this machine did not reproduce the original body-observer failure in 5 whole-file runs; the fix follows the diagnosis in the issue (observation too coarse), not a local red reproduction of that exact assertion

Not run: full desktop e2e suite, npm --workspace @maka/desktop test (no production code change)

AI use

Select exactly one:

• No generative tool made a substantive contribution
• Generative tooling made a substantive contribution

Tool(s) and scope:

Grok authored the test change and this PR description.

Checklist

• Tests cover the change and fail without it
• Lint, format, typecheck and the affected suites pass locally

Does this PR entail a change in behavior?

• Yes — described under Summary above
• No

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: 7cd6910e-da76-4235-9bf8-2171be824736

📥 Commits

Reviewing files that changed from the base of the PR and between 1339419 and 81add45.

📒 Files selected for processing (1)

• apps/desktop/e2e/streaming-remount.spec.ts

🚧 Files skipped from review as they are similar to previous changes (1)

• apps/desktop/e2e/streaming-remount.spec.ts

Included review availability: Your plan includes up to 3 reviews per rolling hour; 1 remains after this review.

---

📝 Walkthrough

Summary

• Fixes intermittent failures in the desktop background-restore E2E test.
• Samples .maka-bubble-streaming on animation frames instead of all document.body mutations.
• Avoids recording React commit intermediates that never paint.
• Confirms that a painted sample contains the settled background text.
• Confirms that no painted sample contains partially accumulated background output.

Design assessment

The change extends the existing E2E test. It does not create a parallel production path or change public contracts.

The frame sampler, sample deduplication, animation tracking, stoppable lifecycle, and next-frame stop are necessary to observe settled painted output reliably. No safe deletion or simplification is evident without weakening regression coverage.

Validation

• Biome checks were run.
• Repeated isolated and whole-file Playwright runs were performed.
• The full desktop E2E suite and desktop unit tests were not run because production code did not change.
• Required-check status remains unverified because direct current diff and status evidence is unavailable.

Review-relevant risks

No protected-area effect was identified in the current diff.

Walkthrough

The background-restore end-to-end test replaces document-wide mutation observation with a stoppable requestAnimationFrame sampler. It records distinct streaming text, tracks active animations, and stops sampling after the next animation frame.

Changes

Streaming restore test

Layer / File(s)	Summary
Frame-based output sampling apps/desktop/e2e/streaming-remount.spec.ts	The test samples the streaming bubble on animation frames, deduplicates text, tracks unfinished animations, waits for one final frame, stops the sampler, and validates restored output.

Estimated code review effort: 2 (Simple) | ~10 minutes

Merge Risk:⚪ Minimal · up to 81add

This localized test-only change adjusts how remount output is sampled without changing product behavior, and no actionable merge-blocking risk remains after normal checks and review.

Possibly related PRs

• maka-agent/maka-agent#3057: Modifies the same streaming-remount end-to-end test and related output-restoration assertions.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly describes the main change: sampling desktop remount restoration on painted animation frames.
Description check	✅ Passed	The description follows the template, explains the problem and solution, lists verification, documents AI use, and records checklist decisions.
Linked Issues check	✅ Passed	The test changes address issue #3061 by sampling painted frames and requiring settled output without partial catch-up states.
Out of Scope Changes check	✅ Passed	The changes are limited to the related desktop streaming-remount E2E test and its observation logic.
Ai Use Disclosure	✅ Passed	The PR selects generative tooling and names Grok with scope. Both introduced commits contain standalone Generated-by: Grok trailers.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

_{Comment @coderabbitai help to get the list of available commands.}

[coderabbitai]

Actionable comments posted: 1

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: cc9670e7-a884-4981-8b6d-710594b20c29

📥 Commits

Reviewing files that changed from the base of the PR and between 9fbdc99 and 1339419.

📒 Files selected for processing (1)

• apps/desktop/e2e/streaming-remount.spec.ts

Included review availability: Your plan includes up to 3 reviews per rolling hour; 2 remain after this review.

[Astro-Han]

Fast-path merge

This change qualifies for the self-merge fast path: it is low impact and easy to reverse (test-only change sampling the remount restore on painted animation frames instead of every DOM mutation), does not touch the protected areas (Runtime Host execution authority, @maka/eval semantics, public Eval CLI, security, licensing, releases, governance), and passes all required checks.

The latest head was independently reviewed by a read-only subagent (ollama-cloud/deepseek-v4-flash:high) with no open P0/P1/P2 findings. Human contributor @yuhan reviewed the final diff and chose the fast path.

[1625567290]

Summary

The "returning to a live conversation settles output accumulated while away" e2e sampled .maka-bubble-streaming on every document.body mutation. That observer sees React commit intermediates that never paint, so the "no partial catch-up" check depended on how much unrelated shell churn happened during remount. It passed in isolation and failed when the rest of the file had already warmed the machine.

The restore is now sampled on animation frames of the streaming bubble only. A real multi-frame replay would still fail; an intra-frame commit that the user never sees would not. The assertion also requires at least one painted sample to contain the settled background text, so a silent no-op observer cannot pass.

Fixes#3061

Verification

• npx biome check apps/desktop/e2e/streaming-remount.spec.ts — clean
• Isolated: playwright test e2e/streaming-remount.spec.ts --grep "accumulated while away" — pass
• Whole file after the change: 3/3, 3/3, plus one earlier whole-file run where this case still passed and an unrelated sibling (keeps a completed reply after an interrupted turn) timed out waiting for 停止 (pre-existing load flake, not this observer)
• Before the change, this machine did not reproduce the original body-observer failure in 5 whole-file runs; the fix follows the diagnosis in the issue (observation too coarse), not a local red reproduction of that exact assertion

Not run: full desktop e2e suite, npm --workspace @maka/desktop test (no production code change)

AI use

Select exactly one:

• No generative tool made a substantive contribution
• Generative tooling made a substantive contribution

Tool(s) and scope:

Grok authored the test change and this PR description.

Checklist

• Tests cover the change and fail without it
• Lint, format, typecheck and the affected suites pass locally

Does this PR entail a change in behavior?

• Yes — described under Summary above
• No

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: 7cd6910e-da76-4235-9bf8-2171be824736

📥 Commits

Reviewing files that changed from the base of the PR and between 1339419 and 81add45.

📒 Files selected for processing (1)

• apps/desktop/e2e/streaming-remount.spec.ts

🚧 Files skipped from review as they are similar to previous changes (1)

• apps/desktop/e2e/streaming-remount.spec.ts

Included review availability: Your plan includes up to 3 reviews per rolling hour; 1 remains after this review.

---

📝 Walkthrough

Summary

• Fixes intermittent failures in the desktop background-restore E2E test.
• Samples .maka-bubble-streaming on animation frames instead of all document.body mutations.
• Avoids recording React commit intermediates that never paint.
• Confirms that a painted sample contains the settled background text.
• Confirms that no painted sample contains partially accumulated background output.

Design assessment

The change extends the existing E2E test. It does not create a parallel production path or change public contracts.

The frame sampler, sample deduplication, animation tracking, stoppable lifecycle, and next-frame stop are necessary to observe settled painted output reliably. No safe deletion or simplification is evident without weakening regression coverage.

Validation

• Biome checks were run.
• Repeated isolated and whole-file Playwright runs were performed.
• The full desktop E2E suite and desktop unit tests were not run because production code did not change.
• Required-check status remains unverified because direct current diff and status evidence is unavailable.

Review-relevant risks

No protected-area effect was identified in the current diff.

Walkthrough

The background-restore end-to-end test replaces document-wide mutation observation with a stoppable requestAnimationFrame sampler. It records distinct streaming text, tracks active animations, and stops sampling after the next animation frame.

Changes

Streaming restore test

Layer / File(s)	Summary
Frame-based output sampling apps/desktop/e2e/streaming-remount.spec.ts	The test samples the streaming bubble on animation frames, deduplicates text, tracks unfinished animations, waits for one final frame, stops the sampler, and validates restored output.

Estimated code review effort: 2 (Simple) | ~10 minutes

Merge Risk:⚪ Minimal · up to 81add

This localized test-only change adjusts how remount output is sampled without changing product behavior, and no actionable merge-blocking risk remains after normal checks and review.

Possibly related PRs

• maka-agent/maka-agent#3057: Modifies the same streaming-remount end-to-end test and related output-restoration assertions.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly describes the main change: sampling desktop remount restoration on painted animation frames.
Description check	✅ Passed	The description follows the template, explains the problem and solution, lists verification, documents AI use, and records checklist decisions.
Linked Issues check	✅ Passed	The test changes address issue #3061 by sampling painted frames and requiring settled output without partial catch-up states.
Out of Scope Changes check	✅ Passed	The changes are limited to the related desktop streaming-remount E2E test and its observation logic.
Ai Use Disclosure	✅ Passed	The PR selects generative tooling and names Grok with scope. Both introduced commits contain standalone Generated-by: Grok trailers.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

_{Comment @coderabbitai help to get the list of available commands.}

[coderabbitai]

Actionable comments posted: 1

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: cc9670e7-a884-4981-8b6d-710594b20c29

📥 Commits

Reviewing files that changed from the base of the PR and between 9fbdc99 and 1339419.

📒 Files selected for processing (1)

• apps/desktop/e2e/streaming-remount.spec.ts

Included review availability: Your plan includes up to 3 reviews per rolling hour; 2 remain after this review.

[Astro-Han]

Fast-path merge

This change qualifies for the self-merge fast path: it is low impact and easy to reverse (test-only change sampling the remount restore on painted animation frames instead of every DOM mutation), does not touch the protected areas (Runtime Host execution authority, @maka/eval semantics, public Eval CLI, security, licensing, releases, governance), and passes all required checks.

The latest head was independently reviewed by a read-only subagent (ollama-cloud/deepseek-v4-flash:high) with no open P0/P1/P2 findings. Human contributor @yuhan reviewed the final diff and chose the fast path.

[1625567290]

Summary

The "returning to a live conversation settles output accumulated while away" e2e sampled .maka-bubble-streaming on every document.body mutation. That observer sees React commit intermediates that never paint, so the "no partial catch-up" check depended on how much unrelated shell churn happened during remount. It passed in isolation and failed when the rest of the file had already warmed the machine.

The restore is now sampled on animation frames of the streaming bubble only. A real multi-frame replay would still fail; an intra-frame commit that the user never sees would not. The assertion also requires at least one painted sample to contain the settled background text, so a silent no-op observer cannot pass.

Fixes#3061

Verification

• npx biome check apps/desktop/e2e/streaming-remount.spec.ts — clean
• Isolated: playwright test e2e/streaming-remount.spec.ts --grep "accumulated while away" — pass
• Whole file after the change: 3/3, 3/3, plus one earlier whole-file run where this case still passed and an unrelated sibling (keeps a completed reply after an interrupted turn) timed out waiting for 停止 (pre-existing load flake, not this observer)
• Before the change, this machine did not reproduce the original body-observer failure in 5 whole-file runs; the fix follows the diagnosis in the issue (observation too coarse), not a local red reproduction of that exact assertion

Not run: full desktop e2e suite, npm --workspace @maka/desktop test (no production code change)

AI use

Select exactly one:

• No generative tool made a substantive contribution
• Generative tooling made a substantive contribution

Tool(s) and scope:

Grok authored the test change and this PR description.

Checklist

• Tests cover the change and fail without it
• Lint, format, typecheck and the affected suites pass locally

Does this PR entail a change in behavior?

• Yes — described under Summary above
• No

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: 7cd6910e-da76-4235-9bf8-2171be824736

📥 Commits

Reviewing files that changed from the base of the PR and between 1339419 and 81add45.

📒 Files selected for processing (1)

• apps/desktop/e2e/streaming-remount.spec.ts

🚧 Files skipped from review as they are similar to previous changes (1)

• apps/desktop/e2e/streaming-remount.spec.ts

Included review availability: Your plan includes up to 3 reviews per rolling hour; 1 remains after this review.

---

📝 Walkthrough

Summary

• Fixes intermittent failures in the desktop background-restore E2E test.
• Samples .maka-bubble-streaming on animation frames instead of all document.body mutations.
• Avoids recording React commit intermediates that never paint.
• Confirms that a painted sample contains the settled background text.
• Confirms that no painted sample contains partially accumulated background output.

Design assessment

The change extends the existing E2E test. It does not create a parallel production path or change public contracts.

The frame sampler, sample deduplication, animation tracking, stoppable lifecycle, and next-frame stop are necessary to observe settled painted output reliably. No safe deletion or simplification is evident without weakening regression coverage.

Validation

• Biome checks were run.
• Repeated isolated and whole-file Playwright runs were performed.
• The full desktop E2E suite and desktop unit tests were not run because production code did not change.
• Required-check status remains unverified because direct current diff and status evidence is unavailable.

Review-relevant risks

No protected-area effect was identified in the current diff.

Walkthrough

The background-restore end-to-end test replaces document-wide mutation observation with a stoppable requestAnimationFrame sampler. It records distinct streaming text, tracks active animations, and stops sampling after the next animation frame.

Changes

Streaming restore test

Layer / File(s)	Summary
Frame-based output sampling apps/desktop/e2e/streaming-remount.spec.ts	The test samples the streaming bubble on animation frames, deduplicates text, tracks unfinished animations, waits for one final frame, stops the sampler, and validates restored output.

Estimated code review effort: 2 (Simple) | ~10 minutes

Merge Risk:⚪ Minimal · up to 81add

This localized test-only change adjusts how remount output is sampled without changing product behavior, and no actionable merge-blocking risk remains after normal checks and review.

Possibly related PRs

• maka-agent/maka-agent#3057: Modifies the same streaming-remount end-to-end test and related output-restoration assertions.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly describes the main change: sampling desktop remount restoration on painted animation frames.
Description check	✅ Passed	The description follows the template, explains the problem and solution, lists verification, documents AI use, and records checklist decisions.
Linked Issues check	✅ Passed	The test changes address issue #3061 by sampling painted frames and requiring settled output without partial catch-up states.
Out of Scope Changes check	✅ Passed	The changes are limited to the related desktop streaming-remount E2E test and its observation logic.
Ai Use Disclosure	✅ Passed	The PR selects generative tooling and names Grok with scope. Both introduced commits contain standalone Generated-by: Grok trailers.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

_{Comment @coderabbitai help to get the list of available commands.}

[coderabbitai]

Actionable comments posted: 1

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: cc9670e7-a884-4981-8b6d-710594b20c29

📥 Commits

Reviewing files that changed from the base of the PR and between 9fbdc99 and 1339419.

📒 Files selected for processing (1)

• apps/desktop/e2e/streaming-remount.spec.ts

Included review availability: Your plan includes up to 3 reviews per rolling hour; 2 remain after this review.

[Astro-Han]

Fast-path merge

This change qualifies for the self-merge fast path: it is low impact and easy to reverse (test-only change sampling the remount restore on painted animation frames instead of every DOM mutation), does not touch the protected areas (Runtime Host execution authority, @maka/eval semantics, public Eval CLI, security, licensing, releases, governance), and passes all required checks.

The latest head was independently reviewed by a read-only subagent (ollama-cloud/deepseek-v4-flash:high) with no open P0/P1/P2 findings. Human contributor @yuhan reviewed the final diff and chose the fast path.