refactor: use official Astryx reasoning by Astro-Han · Pull Request #1748 · apache/maka

Astro-Han · 2026-08-01T04:39:06Z

Summary

replace Maka's custom reasoning disclosure with the official Astryx 0.1.9 ChatReasoning DOM, state machine, keyboard behavior, collapsed preview, chevron, and streaming shimmer
use Astryx's supported swizzle/eject seam because @astryxdesign/lab is canary-only and declares an incompatible exact canary peer for the otherwise identical stable core source; retain the official compiled StyleX atoms without forcing or widening the dependency tree
keep only Maka-owned product behavior around the component: pre-render redaction and hard caps, smooth stream pacing, clipboard feedback through the existing Astryx IconButton, and a textual truncated state
constrain the native component through its public className seam so the official ellipsis shrinks correctly at 520px instead of expanding the 400px conversation column

Verification

npm ci --ignore-scripts
npm --workspace @maka/ui test — 295 passed
npm --workspace @maka/desktop run typecheck
npm --workspace @maka/desktop run build
real Electron fixture: wide collapsed, wide expanded via Enter, and 520×820 expanded capture; the reasoning root is 400px wide with scrollWidth === clientWidth === 400
git diff --check

Visual evidence

Review focus

Astryx publishes ChatReasoning only from @astryxdesign/lab@0.1.9-canary.c9fe437, whose exact core peer conflicts with Maka's stable @astryxdesign/core@0.1.9 even though both point at the v0.1.9 source commit. This PR does not use --force, install the experimental package, or create a compatibility wrapper. It ejects that one official component with provenance and license headers, using the byte-equivalent compiled atoms from the published component. Core already ships all but the five official shimmer atoms; those five declarations are retained verbatim beside the chat CSS.

Astryx has no redaction, stream-buffer pacing, whole-message clipboard lifecycle, or security truncation component. Those remain Maka product responsibilities. Visual truncation of the collapsed preview and the streaming label treatment are owned by ChatReasoning.

…1938) * fix(ui): restore line wrapping in the deep-thinking disclosure body The ejected Astryx ChatReasoning content shell owns no white-space: its StyleX atoms declare only padding/typography/color, so the inherited white-space: normal collapsed every newline in thinking text, swallowing the reasoning line structure (regression from the #1748 migration, which replaced the old whitespace-pre-wrap body with the official component). Give the reasoning body a product class (maka-chat-reasoning-content) on the eject seam and restore the pre-wrap reading contract in @maka/ui styles.css, matching the tool-output body's existing pattern. Regression coverage: - ui render test locks the class on the content div - desktop CSS contract asserts the pre-wrap rule is the only declaration on that class (fails red without the rule) * test(desktop): assert final effective wrap value, not a rule count Codex review (P3): the contract test's 'exactly once' count was a proxy for the real invariant and over-constrained — a future harmless second rule (e.g. a focus outline, a media variant) would false-fail even though white-space stayed pre-wrap. Walk every matching rule body in cascade order and assert the LAST declaration of white-space/word-break is pre-wrap/break-word. Verified: a later white-space: normal re-declaration fails red; a harmless margin addition stays green; the single rule passes.

Astro-Han · 2026-08-01T04:39:06Z

Summary

replace Maka's custom reasoning disclosure with the official Astryx 0.1.9 ChatReasoning DOM, state machine, keyboard behavior, collapsed preview, chevron, and streaming shimmer
use Astryx's supported swizzle/eject seam because @astryxdesign/lab is canary-only and declares an incompatible exact canary peer for the otherwise identical stable core source; retain the official compiled StyleX atoms without forcing or widening the dependency tree
keep only Maka-owned product behavior around the component: pre-render redaction and hard caps, smooth stream pacing, clipboard feedback through the existing Astryx IconButton, and a textual truncated state
constrain the native component through its public className seam so the official ellipsis shrinks correctly at 520px instead of expanding the 400px conversation column

Verification

npm ci --ignore-scripts
npm --workspace @maka/ui test — 295 passed
npm --workspace @maka/desktop run typecheck
npm --workspace @maka/desktop run build
real Electron fixture: wide collapsed, wide expanded via Enter, and 520×820 expanded capture; the reasoning root is 400px wide with scrollWidth === clientWidth === 400
git diff --check

Visual evidence

Review focus

Astryx publishes ChatReasoning only from @astryxdesign/lab@0.1.9-canary.c9fe437, whose exact core peer conflicts with Maka's stable @astryxdesign/core@0.1.9 even though both point at the v0.1.9 source commit. This PR does not use --force, install the experimental package, or create a compatibility wrapper. It ejects that one official component with provenance and license headers, using the byte-equivalent compiled atoms from the published component. Core already ships all but the five official shimmer atoms; those five declarations are retained verbatim beside the chat CSS.

Astryx has no redaction, stream-buffer pacing, whole-message clipboard lifecycle, or security truncation component. Those remain Maka product responsibilities. Visual truncation of the collapsed preview and the streaming label treatment are owned by ChatReasoning.

…1938) * fix(ui): restore line wrapping in the deep-thinking disclosure body The ejected Astryx ChatReasoning content shell owns no white-space: its StyleX atoms declare only padding/typography/color, so the inherited white-space: normal collapsed every newline in thinking text, swallowing the reasoning line structure (regression from the #1748 migration, which replaced the old whitespace-pre-wrap body with the official component). Give the reasoning body a product class (maka-chat-reasoning-content) on the eject seam and restore the pre-wrap reading contract in @maka/ui styles.css, matching the tool-output body's existing pattern. Regression coverage: - ui render test locks the class on the content div - desktop CSS contract asserts the pre-wrap rule is the only declaration on that class (fails red without the rule) * test(desktop): assert final effective wrap value, not a rule count Codex review (P3): the contract test's 'exactly once' count was a proxy for the real invariant and over-constrained — a future harmless second rule (e.g. a focus outline, a media variant) would false-fail even though white-space stayed pre-wrap. Walk every matching rule body in cascade order and assert the LAST declaration of white-space/word-break is pre-wrap/break-word. Verified: a later white-space: normal re-declaration fails red; a harmless margin addition stays green; the single rule passes.

Astro-Han · 2026-08-01T04:39:06Z

Summary

replace Maka's custom reasoning disclosure with the official Astryx 0.1.9 ChatReasoning DOM, state machine, keyboard behavior, collapsed preview, chevron, and streaming shimmer
use Astryx's supported swizzle/eject seam because @astryxdesign/lab is canary-only and declares an incompatible exact canary peer for the otherwise identical stable core source; retain the official compiled StyleX atoms without forcing or widening the dependency tree
keep only Maka-owned product behavior around the component: pre-render redaction and hard caps, smooth stream pacing, clipboard feedback through the existing Astryx IconButton, and a textual truncated state
constrain the native component through its public className seam so the official ellipsis shrinks correctly at 520px instead of expanding the 400px conversation column

Verification

npm ci --ignore-scripts
npm --workspace @maka/ui test — 295 passed
npm --workspace @maka/desktop run typecheck
npm --workspace @maka/desktop run build
real Electron fixture: wide collapsed, wide expanded via Enter, and 520×820 expanded capture; the reasoning root is 400px wide with scrollWidth === clientWidth === 400
git diff --check

Visual evidence

Review focus

Astryx publishes ChatReasoning only from @astryxdesign/lab@0.1.9-canary.c9fe437, whose exact core peer conflicts with Maka's stable @astryxdesign/core@0.1.9 even though both point at the v0.1.9 source commit. This PR does not use --force, install the experimental package, or create a compatibility wrapper. It ejects that one official component with provenance and license headers, using the byte-equivalent compiled atoms from the published component. Core already ships all but the five official shimmer atoms; those five declarations are retained verbatim beside the chat CSS.

Astryx has no redaction, stream-buffer pacing, whole-message clipboard lifecycle, or security truncation component. Those remain Maka product responsibilities. Visual truncation of the collapsed preview and the streaming label treatment are owned by ChatReasoning.

…1938) * fix(ui): restore line wrapping in the deep-thinking disclosure body The ejected Astryx ChatReasoning content shell owns no white-space: its StyleX atoms declare only padding/typography/color, so the inherited white-space: normal collapsed every newline in thinking text, swallowing the reasoning line structure (regression from the #1748 migration, which replaced the old whitespace-pre-wrap body with the official component). Give the reasoning body a product class (maka-chat-reasoning-content) on the eject seam and restore the pre-wrap reading contract in @maka/ui styles.css, matching the tool-output body's existing pattern. Regression coverage: - ui render test locks the class on the content div - desktop CSS contract asserts the pre-wrap rule is the only declaration on that class (fails red without the rule) * test(desktop): assert final effective wrap value, not a rule count Codex review (P3): the contract test's 'exactly once' count was a proxy for the real invariant and over-constrained — a future harmless second rule (e.g. a focus outline, a media variant) would false-fail even though white-space stayed pre-wrap. Walk every matching rule body in cascade order and assert the LAST declaration of white-space/word-break is pre-wrap/break-word. Verified: a later white-space: normal re-declaration fails red; a harmless margin addition stays green; the single rule passes.

Astro-Han · 2026-08-01T04:39:06Z

Summary

replace Maka's custom reasoning disclosure with the official Astryx 0.1.9 ChatReasoning DOM, state machine, keyboard behavior, collapsed preview, chevron, and streaming shimmer
use Astryx's supported swizzle/eject seam because @astryxdesign/lab is canary-only and declares an incompatible exact canary peer for the otherwise identical stable core source; retain the official compiled StyleX atoms without forcing or widening the dependency tree
keep only Maka-owned product behavior around the component: pre-render redaction and hard caps, smooth stream pacing, clipboard feedback through the existing Astryx IconButton, and a textual truncated state
constrain the native component through its public className seam so the official ellipsis shrinks correctly at 520px instead of expanding the 400px conversation column

Verification

npm ci --ignore-scripts
npm --workspace @maka/ui test — 295 passed
npm --workspace @maka/desktop run typecheck
npm --workspace @maka/desktop run build
real Electron fixture: wide collapsed, wide expanded via Enter, and 520×820 expanded capture; the reasoning root is 400px wide with scrollWidth === clientWidth === 400
git diff --check

Visual evidence

Review focus

Astryx publishes ChatReasoning only from @astryxdesign/lab@0.1.9-canary.c9fe437, whose exact core peer conflicts with Maka's stable @astryxdesign/core@0.1.9 even though both point at the v0.1.9 source commit. This PR does not use --force, install the experimental package, or create a compatibility wrapper. It ejects that one official component with provenance and license headers, using the byte-equivalent compiled atoms from the published component. Core already ships all but the five official shimmer atoms; those five declarations are retained verbatim beside the chat CSS.

Astryx has no redaction, stream-buffer pacing, whole-message clipboard lifecycle, or security truncation component. Those remain Maka product responsibilities. Visual truncation of the collapsed preview and the streaming label treatment are owned by ChatReasoning.

…1938) * fix(ui): restore line wrapping in the deep-thinking disclosure body The ejected Astryx ChatReasoning content shell owns no white-space: its StyleX atoms declare only padding/typography/color, so the inherited white-space: normal collapsed every newline in thinking text, swallowing the reasoning line structure (regression from the #1748 migration, which replaced the old whitespace-pre-wrap body with the official component). Give the reasoning body a product class (maka-chat-reasoning-content) on the eject seam and restore the pre-wrap reading contract in @maka/ui styles.css, matching the tool-output body's existing pattern. Regression coverage: - ui render test locks the class on the content div - desktop CSS contract asserts the pre-wrap rule is the only declaration on that class (fails red without the rule) * test(desktop): assert final effective wrap value, not a rule count Codex review (P3): the contract test's 'exactly once' count was a proxy for the real invariant and over-constrained — a future harmless second rule (e.g. a focus outline, a media variant) would false-fail even though white-space stayed pre-wrap. Walk every matching rule body in cascade order and assert the LAST declaration of white-space/word-break is pre-wrap/break-word. Verified: a later white-space: normal re-declaration fails red; a harmless margin addition stays green; the single rule passes.

Astro-Han · 2026-08-01T04:39:06Z

Summary

replace Maka's custom reasoning disclosure with the official Astryx 0.1.9 ChatReasoning DOM, state machine, keyboard behavior, collapsed preview, chevron, and streaming shimmer
use Astryx's supported swizzle/eject seam because @astryxdesign/lab is canary-only and declares an incompatible exact canary peer for the otherwise identical stable core source; retain the official compiled StyleX atoms without forcing or widening the dependency tree
keep only Maka-owned product behavior around the component: pre-render redaction and hard caps, smooth stream pacing, clipboard feedback through the existing Astryx IconButton, and a textual truncated state
constrain the native component through its public className seam so the official ellipsis shrinks correctly at 520px instead of expanding the 400px conversation column

Verification

npm ci --ignore-scripts
npm --workspace @maka/ui test — 295 passed
npm --workspace @maka/desktop run typecheck
npm --workspace @maka/desktop run build
real Electron fixture: wide collapsed, wide expanded via Enter, and 520×820 expanded capture; the reasoning root is 400px wide with scrollWidth === clientWidth === 400
git diff --check

Visual evidence

Review focus

Astryx publishes ChatReasoning only from @astryxdesign/lab@0.1.9-canary.c9fe437, whose exact core peer conflicts with Maka's stable @astryxdesign/core@0.1.9 even though both point at the v0.1.9 source commit. This PR does not use --force, install the experimental package, or create a compatibility wrapper. It ejects that one official component with provenance and license headers, using the byte-equivalent compiled atoms from the published component. Core already ships all but the five official shimmer atoms; those five declarations are retained verbatim beside the chat CSS.

Astryx has no redaction, stream-buffer pacing, whole-message clipboard lifecycle, or security truncation component. Those remain Maka product responsibilities. Visual truncation of the collapsed preview and the streaming label treatment are owned by ChatReasoning.

…1938) * fix(ui): restore line wrapping in the deep-thinking disclosure body The ejected Astryx ChatReasoning content shell owns no white-space: its StyleX atoms declare only padding/typography/color, so the inherited white-space: normal collapsed every newline in thinking text, swallowing the reasoning line structure (regression from the #1748 migration, which replaced the old whitespace-pre-wrap body with the official component). Give the reasoning body a product class (maka-chat-reasoning-content) on the eject seam and restore the pre-wrap reading contract in @maka/ui styles.css, matching the tool-output body's existing pattern. Regression coverage: - ui render test locks the class on the content div - desktop CSS contract asserts the pre-wrap rule is the only declaration on that class (fails red without the rule) * test(desktop): assert final effective wrap value, not a rule count Codex review (P3): the contract test's 'exactly once' count was a proxy for the real invariant and over-constrained — a future harmless second rule (e.g. a focus outline, a media variant) would false-fail even though white-space stayed pre-wrap. Walk every matching rule body in cascade order and assert the LAST declaration of white-space/word-break is pre-wrap/break-word. Verified: a later white-space: normal re-declaration fails red; a harmless margin addition stays green; the single rule passes.

Astro-Han · 2026-08-01T04:39:06Z

Summary

replace Maka's custom reasoning disclosure with the official Astryx 0.1.9 ChatReasoning DOM, state machine, keyboard behavior, collapsed preview, chevron, and streaming shimmer
use Astryx's supported swizzle/eject seam because @astryxdesign/lab is canary-only and declares an incompatible exact canary peer for the otherwise identical stable core source; retain the official compiled StyleX atoms without forcing or widening the dependency tree
keep only Maka-owned product behavior around the component: pre-render redaction and hard caps, smooth stream pacing, clipboard feedback through the existing Astryx IconButton, and a textual truncated state
constrain the native component through its public className seam so the official ellipsis shrinks correctly at 520px instead of expanding the 400px conversation column

Verification

npm ci --ignore-scripts
npm --workspace @maka/ui test — 295 passed
npm --workspace @maka/desktop run typecheck
npm --workspace @maka/desktop run build
real Electron fixture: wide collapsed, wide expanded via Enter, and 520×820 expanded capture; the reasoning root is 400px wide with scrollWidth === clientWidth === 400
git diff --check

Visual evidence

Review focus

Astryx publishes ChatReasoning only from @astryxdesign/lab@0.1.9-canary.c9fe437, whose exact core peer conflicts with Maka's stable @astryxdesign/core@0.1.9 even though both point at the v0.1.9 source commit. This PR does not use --force, install the experimental package, or create a compatibility wrapper. It ejects that one official component with provenance and license headers, using the byte-equivalent compiled atoms from the published component. Core already ships all but the five official shimmer atoms; those five declarations are retained verbatim beside the chat CSS.

Astryx has no redaction, stream-buffer pacing, whole-message clipboard lifecycle, or security truncation component. Those remain Maka product responsibilities. Visual truncation of the collapsed preview and the streaming label treatment are owned by ChatReasoning.

…1938) * fix(ui): restore line wrapping in the deep-thinking disclosure body The ejected Astryx ChatReasoning content shell owns no white-space: its StyleX atoms declare only padding/typography/color, so the inherited white-space: normal collapsed every newline in thinking text, swallowing the reasoning line structure (regression from the #1748 migration, which replaced the old whitespace-pre-wrap body with the official component). Give the reasoning body a product class (maka-chat-reasoning-content) on the eject seam and restore the pre-wrap reading contract in @maka/ui styles.css, matching the tool-output body's existing pattern. Regression coverage: - ui render test locks the class on the content div - desktop CSS contract asserts the pre-wrap rule is the only declaration on that class (fails red without the rule) * test(desktop): assert final effective wrap value, not a rule count Codex review (P3): the contract test's 'exactly once' count was a proxy for the real invariant and over-constrained — a future harmless second rule (e.g. a focus outline, a media variant) would false-fail even though white-space stayed pre-wrap. Walk every matching rule body in cascade order and assert the LAST declaration of white-space/word-break is pre-wrap/break-word. Verified: a later white-space: normal re-declaration fails red; a harmless margin addition stays green; the single rule passes.

Astro-Han · 2026-08-01T04:39:06Z

Summary

replace Maka's custom reasoning disclosure with the official Astryx 0.1.9 ChatReasoning DOM, state machine, keyboard behavior, collapsed preview, chevron, and streaming shimmer
use Astryx's supported swizzle/eject seam because @astryxdesign/lab is canary-only and declares an incompatible exact canary peer for the otherwise identical stable core source; retain the official compiled StyleX atoms without forcing or widening the dependency tree
keep only Maka-owned product behavior around the component: pre-render redaction and hard caps, smooth stream pacing, clipboard feedback through the existing Astryx IconButton, and a textual truncated state
constrain the native component through its public className seam so the official ellipsis shrinks correctly at 520px instead of expanding the 400px conversation column

Verification

npm ci --ignore-scripts
npm --workspace @maka/ui test — 295 passed
npm --workspace @maka/desktop run typecheck
npm --workspace @maka/desktop run build
real Electron fixture: wide collapsed, wide expanded via Enter, and 520×820 expanded capture; the reasoning root is 400px wide with scrollWidth === clientWidth === 400
git diff --check

Visual evidence

Review focus

Astryx publishes ChatReasoning only from @astryxdesign/lab@0.1.9-canary.c9fe437, whose exact core peer conflicts with Maka's stable @astryxdesign/core@0.1.9 even though both point at the v0.1.9 source commit. This PR does not use --force, install the experimental package, or create a compatibility wrapper. It ejects that one official component with provenance and license headers, using the byte-equivalent compiled atoms from the published component. Core already ships all but the five official shimmer atoms; those five declarations are retained verbatim beside the chat CSS.

Astryx has no redaction, stream-buffer pacing, whole-message clipboard lifecycle, or security truncation component. Those remain Maka product responsibilities. Visual truncation of the collapsed preview and the streaming label treatment are owned by ChatReasoning.

…1938) * fix(ui): restore line wrapping in the deep-thinking disclosure body The ejected Astryx ChatReasoning content shell owns no white-space: its StyleX atoms declare only padding/typography/color, so the inherited white-space: normal collapsed every newline in thinking text, swallowing the reasoning line structure (regression from the #1748 migration, which replaced the old whitespace-pre-wrap body with the official component). Give the reasoning body a product class (maka-chat-reasoning-content) on the eject seam and restore the pre-wrap reading contract in @maka/ui styles.css, matching the tool-output body's existing pattern. Regression coverage: - ui render test locks the class on the content div - desktop CSS contract asserts the pre-wrap rule is the only declaration on that class (fails red without the rule) * test(desktop): assert final effective wrap value, not a rule count Codex review (P3): the contract test's 'exactly once' count was a proxy for the real invariant and over-constrained — a future harmless second rule (e.g. a focus outline, a media variant) would false-fail even though white-space stayed pre-wrap. Walk every matching rule body in cascade order and assert the LAST declaration of white-space/word-break is pre-wrap/break-word. Verified: a later white-space: normal re-declaration fails red; a harmless margin addition stays green; the single rule passes.

Astro-Han · 2026-08-01T04:39:06Z

Summary

replace Maka's custom reasoning disclosure with the official Astryx 0.1.9 ChatReasoning DOM, state machine, keyboard behavior, collapsed preview, chevron, and streaming shimmer
use Astryx's supported swizzle/eject seam because @astryxdesign/lab is canary-only and declares an incompatible exact canary peer for the otherwise identical stable core source; retain the official compiled StyleX atoms without forcing or widening the dependency tree
keep only Maka-owned product behavior around the component: pre-render redaction and hard caps, smooth stream pacing, clipboard feedback through the existing Astryx IconButton, and a textual truncated state
constrain the native component through its public className seam so the official ellipsis shrinks correctly at 520px instead of expanding the 400px conversation column

Verification

npm ci --ignore-scripts
npm --workspace @maka/ui test — 295 passed
npm --workspace @maka/desktop run typecheck
npm --workspace @maka/desktop run build
real Electron fixture: wide collapsed, wide expanded via Enter, and 520×820 expanded capture; the reasoning root is 400px wide with scrollWidth === clientWidth === 400
git diff --check

Visual evidence

Review focus

Astryx publishes ChatReasoning only from @astryxdesign/lab@0.1.9-canary.c9fe437, whose exact core peer conflicts with Maka's stable @astryxdesign/core@0.1.9 even though both point at the v0.1.9 source commit. This PR does not use --force, install the experimental package, or create a compatibility wrapper. It ejects that one official component with provenance and license headers, using the byte-equivalent compiled atoms from the published component. Core already ships all but the five official shimmer atoms; those five declarations are retained verbatim beside the chat CSS.

Astryx has no redaction, stream-buffer pacing, whole-message clipboard lifecycle, or security truncation component. Those remain Maka product responsibilities. Visual truncation of the collapsed preview and the streaming label treatment are owned by ChatReasoning.

…1938) * fix(ui): restore line wrapping in the deep-thinking disclosure body The ejected Astryx ChatReasoning content shell owns no white-space: its StyleX atoms declare only padding/typography/color, so the inherited white-space: normal collapsed every newline in thinking text, swallowing the reasoning line structure (regression from the #1748 migration, which replaced the old whitespace-pre-wrap body with the official component). Give the reasoning body a product class (maka-chat-reasoning-content) on the eject seam and restore the pre-wrap reading contract in @maka/ui styles.css, matching the tool-output body's existing pattern. Regression coverage: - ui render test locks the class on the content div - desktop CSS contract asserts the pre-wrap rule is the only declaration on that class (fails red without the rule) * test(desktop): assert final effective wrap value, not a rule count Codex review (P3): the contract test's 'exactly once' count was a proxy for the real invariant and over-constrained — a future harmless second rule (e.g. a focus outline, a media variant) would false-fail even though white-space stayed pre-wrap. Walk every matching rule body in cascade order and assert the LAST declaration of white-space/word-break is pre-wrap/break-word. Verified: a later white-space: normal re-declaration fails red; a harmless margin addition stays green; the single rule passes.

Astro-Han added 2 commits August 1, 2026 12:37

refactor: use official Astryx reasoning
43282da

refactor: remove obsolete processing summaries
2152b13

Astro-Han merged commit 82f7dfc into mainAug 1, 2026
2 of 3 checks passed

Astro-Han deleted the feat/slice9-astryx-conversation branch August 1, 2026 04:57

Astro-Han mentioned this pull request Aug 3, 2026
fix(ui): restore line wrapping in the deep-thinking disclosure body #1938
Merged

Astro-Han mentioned this pull request Aug 24, 2026
fix: surface and render OpenAI and Claude reasoning summaries #2645
Merged
6 tasks

Astro-Han added 2 commits August 1, 2026 12:37

refactor: use official Astryx reasoning
43282da

refactor: remove obsolete processing summaries
2152b13

Astro-Han merged commit 82f7dfc into mainAug 1, 2026
2 of 3 checks passed

Astro-Han deleted the feat/slice9-astryx-conversation branch August 1, 2026 04:57

Astro-Han mentioned this pull request Aug 3, 2026
fix(ui): restore line wrapping in the deep-thinking disclosure body #1938
Merged

Astro-Han mentioned this pull request Aug 24, 2026
fix: surface and render OpenAI and Claude reasoning summaries #2645
Merged
6 tasks

Astro-Han added 2 commits August 1, 2026 12:37

refactor: use official Astryx reasoning
43282da

refactor: remove obsolete processing summaries
2152b13

Astro-Han merged commit 82f7dfc into mainAug 1, 2026
2 of 3 checks passed

Astro-Han deleted the feat/slice9-astryx-conversation branch August 1, 2026 04:57

Astro-Han mentioned this pull request Aug 3, 2026
fix(ui): restore line wrapping in the deep-thinking disclosure body #1938
Merged

Astro-Han mentioned this pull request Aug 24, 2026
fix: surface and render OpenAI and Claude reasoning summaries #2645
Merged
6 tasks

Astro-Han added 2 commits August 1, 2026 12:37

refactor: use official Astryx reasoning
43282da

refactor: remove obsolete processing summaries
2152b13

Astro-Han merged commit 82f7dfc into mainAug 1, 2026
2 of 3 checks passed

Astro-Han deleted the feat/slice9-astryx-conversation branch August 1, 2026 04:57

Astro-Han mentioned this pull request Aug 3, 2026
fix(ui): restore line wrapping in the deep-thinking disclosure body #1938
Merged

Astro-Han mentioned this pull request Aug 24, 2026
fix: surface and render OpenAI and Claude reasoning summaries #2645
Merged
6 tasks

Astro-Han added 2 commits August 1, 2026 12:37

refactor: use official Astryx reasoning
43282da

refactor: remove obsolete processing summaries
2152b13

Astro-Han merged commit 82f7dfc into mainAug 1, 2026
2 of 3 checks passed

Astro-Han deleted the feat/slice9-astryx-conversation branch August 1, 2026 04:57

Astro-Han mentioned this pull request Aug 3, 2026
fix(ui): restore line wrapping in the deep-thinking disclosure body #1938
Merged

Astro-Han mentioned this pull request Aug 24, 2026
fix: surface and render OpenAI and Claude reasoning summaries #2645
Merged
6 tasks

Astro-Han added 2 commits August 1, 2026 12:37

refactor: use official Astryx reasoning
43282da

refactor: remove obsolete processing summaries
2152b13

Astro-Han merged commit 82f7dfc into mainAug 1, 2026
2 of 3 checks passed

Astro-Han deleted the feat/slice9-astryx-conversation branch August 1, 2026 04:57

Astro-Han mentioned this pull request Aug 3, 2026
fix(ui): restore line wrapping in the deep-thinking disclosure body #1938
Merged

Astro-Han mentioned this pull request Aug 24, 2026
fix: surface and render OpenAI and Claude reasoning summaries #2645
Merged
6 tasks

Astro-Han added 2 commits August 1, 2026 12:37

refactor: use official Astryx reasoning
43282da

refactor: remove obsolete processing summaries
2152b13

Astro-Han merged commit 82f7dfc into mainAug 1, 2026
2 of 3 checks passed

Astro-Han deleted the feat/slice9-astryx-conversation branch August 1, 2026 04:57

Astro-Han mentioned this pull request Aug 3, 2026
fix(ui): restore line wrapping in the deep-thinking disclosure body #1938
Merged

Astro-Han mentioned this pull request Aug 24, 2026
fix: surface and render OpenAI and Claude reasoning summaries #2645
Merged
6 tasks

Astro-Han added 2 commits August 1, 2026 12:37

refactor: use official Astryx reasoning
43282da

refactor: remove obsolete processing summaries
2152b13

Astro-Han merged commit 82f7dfc into mainAug 1, 2026
2 of 3 checks passed

Astro-Han deleted the feat/slice9-astryx-conversation branch August 1, 2026 04:57

Astro-Han mentioned this pull request Aug 3, 2026
fix(ui): restore line wrapping in the deep-thinking disclosure body #1938
Merged

Astro-Han mentioned this pull request Aug 24, 2026
fix: surface and render OpenAI and Claude reasoning summaries #2645
Merged
6 tasks

Uh oh!

Conversation

Astro-Han commented Aug 1, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Verification

Visual evidence

Review focus

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Uh oh!

Conversation

Astro-Han commented Aug 1, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Verification

Visual evidence

Review focus

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Uh oh!

Conversation

Astro-Han commented Aug 1, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Verification

Visual evidence

Review focus

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Uh oh!

Conversation

Astro-Han commented Aug 1, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Verification

Visual evidence

Review focus

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Uh oh!

Conversation

Astro-Han commented Aug 1, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Verification

Visual evidence

Review focus

Astro-Han commented Aug 1, 2026•
edited
Loading

Astro-Han commented Aug 1, 2026•
edited
Loading

Astro-Han commented Aug 1, 2026•
edited
Loading

Astro-Han commented Aug 1, 2026•
edited
Loading

Astro-Han commented Aug 1, 2026•
edited
Loading

Astro-Han commented Aug 1, 2026•
edited
Loading

Astro-Han commented Aug 1, 2026•
edited
Loading