fix(web): key ChatMarkdown renderers on thread ids, not object identity

[juliusmarminge]

Problem

Follow-up to #9306. ChatMarkdown's markdownComponents memo depends on threadRef, and react-markdown uses those renderer functions as React element types. Any caller that passes a fresh-but-equal threadRef object therefore remounts every rendered <p>/<pre>/<code> in the message — which is what was destroying native text selections and dropping the cite toolbar on working threads. #9306 fixed the one caller doing that, but the component stayed one careless parseScopedThreadKey(...) away from regressing.

Fix

Memoize threadRef on environmentId + threadId at the top of ChatMarkdown, before anything downstream depends on it. Every existing use (markdownComponents, the PR-link callbacks, the media effect, the child memo comparators) now sees a stable object as long as the ids are unchanged, regardless of what the caller passes.

Verified with vp lint, apps/web typecheck, and the ChatMarkdown tests (69 passing).

---

Claude Fable 5 via Claude Code

🤖 Generated with Claude Code

---

Note

Low Risk
Small, localized React memoization in chat markdown rendering with no auth, data, or API changes.

Overview
Stabilizes threadRef inside ChatMarkdown so react-markdown does not treat every new object reference as a renderer change and remount the whole message tree (which was clearing native text selection and breaking the cite toolbar).

The prop is renamed to threadRefProp and wrapped in useMemo keyed on environmentId and threadId. Downstream logic (markdownComponents, PR-link handlers, media preview, effects) keeps the same API but sees a stable reference while the scoped thread is unchanged, even if parents pass a freshly allocated ScopedThreadRef each render.

^{Reviewed by Cursor Bugbot for commit 834f807. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Fix ChatMarkdown to key renderers on thread IDs, not object identity

Memoizes the thread reference inside ChatMarkdown using environment and thread IDs so an equivalent-but-new thread object does not create a new renderer identity. This preserves native text selections inside rendered markdown nodes across re-renders when the IDs are unchanged.

^{Macroscope summarized 834f807.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.5 KiB	13.2 KiB	−219 B (−1.6%)	15.1 KiB	✅
Codex	Thread snapshot wire	6.9 KiB	6.9 KiB	+5 B (+0.1%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.6 KiB	6.3 KiB	−224 B (−3.3%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	57.0 KiB	55.5 KiB	−1.5 KiB (−2.6%)	66.4 KiB	✅
Codex	Live turn messages	10	8	−2 (−20.0%)	21	✅
Claude	Total thread wire	13.3 KiB	13.3 KiB	+20 B (+0.1%)	15.1 KiB	✅
Claude	Thread snapshot wire	6.9 KiB	6.9 KiB	+5 B (+0.1%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.4 KiB	6.4 KiB	+15 B (+0.2%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	56.4 KiB	56.4 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	10	10	0 (0.0%)	21	✅

Baseline: 355fbd9 · PR result: 834f807 · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 109.4 KiB
• Claude decoded thread snapshot: 110.1 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.

[macroscopeapp]

Approvability

Verdict: Approved at 834f807

Macroscope's review found this PR approvable — This is a narrowly scoped one-file fix that stabilizes existing ChatMarkdown renderer identity without introducing new capabilities, schema changes, defaults, or sensitive behavior. Renderer updates still occur when the environment or thread ID changes.

^{You can add or adjust custom eligibility rules. Learn more.}

[juliusmarminge]

Problem

Follow-up to #9306. ChatMarkdown's markdownComponents memo depends on threadRef, and react-markdown uses those renderer functions as React element types. Any caller that passes a fresh-but-equal threadRef object therefore remounts every rendered <p>/<pre>/<code> in the message — which is what was destroying native text selections and dropping the cite toolbar on working threads. #9306 fixed the one caller doing that, but the component stayed one careless parseScopedThreadKey(...) away from regressing.

Fix

Memoize threadRef on environmentId + threadId at the top of ChatMarkdown, before anything downstream depends on it. Every existing use (markdownComponents, the PR-link callbacks, the media effect, the child memo comparators) now sees a stable object as long as the ids are unchanged, regardless of what the caller passes.

Verified with vp lint, apps/web typecheck, and the ChatMarkdown tests (69 passing).

---

Claude Fable 5 via Claude Code

🤖 Generated with Claude Code

---

Note

Low Risk
Small, localized React memoization in chat markdown rendering with no auth, data, or API changes.

Overview
Stabilizes threadRef inside ChatMarkdown so react-markdown does not treat every new object reference as a renderer change and remount the whole message tree (which was clearing native text selection and breaking the cite toolbar).

The prop is renamed to threadRefProp and wrapped in useMemo keyed on environmentId and threadId. Downstream logic (markdownComponents, PR-link handlers, media preview, effects) keeps the same API but sees a stable reference while the scoped thread is unchanged, even if parents pass a freshly allocated ScopedThreadRef each render.

^{Reviewed by Cursor Bugbot for commit 834f807. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Fix ChatMarkdown to key renderers on thread IDs, not object identity

Memoizes the thread reference inside ChatMarkdown using environment and thread IDs so an equivalent-but-new thread object does not create a new renderer identity. This preserves native text selections inside rendered markdown nodes across re-renders when the IDs are unchanged.

^{Macroscope summarized 834f807.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.5 KiB	13.2 KiB	−219 B (−1.6%)	15.1 KiB	✅
Codex	Thread snapshot wire	6.9 KiB	6.9 KiB	+5 B (+0.1%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.6 KiB	6.3 KiB	−224 B (−3.3%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	57.0 KiB	55.5 KiB	−1.5 KiB (−2.6%)	66.4 KiB	✅
Codex	Live turn messages	10	8	−2 (−20.0%)	21	✅
Claude	Total thread wire	13.3 KiB	13.3 KiB	+20 B (+0.1%)	15.1 KiB	✅
Claude	Thread snapshot wire	6.9 KiB	6.9 KiB	+5 B (+0.1%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.4 KiB	6.4 KiB	+15 B (+0.2%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	56.4 KiB	56.4 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	10	10	0 (0.0%)	21	✅

Baseline: 355fbd9 · PR result: 834f807 · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 109.4 KiB
• Claude decoded thread snapshot: 110.1 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.

[macroscopeapp]

Approvability

Verdict: Approved at 834f807

Macroscope's review found this PR approvable — This is a narrowly scoped one-file fix that stabilizes existing ChatMarkdown renderer identity without introducing new capabilities, schema changes, defaults, or sensitive behavior. Renderer updates still occur when the environment or thread ID changes.

^{You can add or adjust custom eligibility rules. Learn more.}

[juliusmarminge]

Problem

Follow-up to #9306. ChatMarkdown's markdownComponents memo depends on threadRef, and react-markdown uses those renderer functions as React element types. Any caller that passes a fresh-but-equal threadRef object therefore remounts every rendered <p>/<pre>/<code> in the message — which is what was destroying native text selections and dropping the cite toolbar on working threads. #9306 fixed the one caller doing that, but the component stayed one careless parseScopedThreadKey(...) away from regressing.

Fix

Memoize threadRef on environmentId + threadId at the top of ChatMarkdown, before anything downstream depends on it. Every existing use (markdownComponents, the PR-link callbacks, the media effect, the child memo comparators) now sees a stable object as long as the ids are unchanged, regardless of what the caller passes.

Verified with vp lint, apps/web typecheck, and the ChatMarkdown tests (69 passing).

---

Claude Fable 5 via Claude Code

🤖 Generated with Claude Code

---

Note

Low Risk
Small, localized React memoization in chat markdown rendering with no auth, data, or API changes.

Overview
Stabilizes threadRef inside ChatMarkdown so react-markdown does not treat every new object reference as a renderer change and remount the whole message tree (which was clearing native text selection and breaking the cite toolbar).

The prop is renamed to threadRefProp and wrapped in useMemo keyed on environmentId and threadId. Downstream logic (markdownComponents, PR-link handlers, media preview, effects) keeps the same API but sees a stable reference while the scoped thread is unchanged, even if parents pass a freshly allocated ScopedThreadRef each render.

^{Reviewed by Cursor Bugbot for commit 834f807. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Fix ChatMarkdown to key renderers on thread IDs, not object identity

Memoizes the thread reference inside ChatMarkdown using environment and thread IDs so an equivalent-but-new thread object does not create a new renderer identity. This preserves native text selections inside rendered markdown nodes across re-renders when the IDs are unchanged.

^{Macroscope summarized 834f807.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.5 KiB	13.2 KiB	−219 B (−1.6%)	15.1 KiB	✅
Codex	Thread snapshot wire	6.9 KiB	6.9 KiB	+5 B (+0.1%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.6 KiB	6.3 KiB	−224 B (−3.3%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	57.0 KiB	55.5 KiB	−1.5 KiB (−2.6%)	66.4 KiB	✅
Codex	Live turn messages	10	8	−2 (−20.0%)	21	✅
Claude	Total thread wire	13.3 KiB	13.3 KiB	+20 B (+0.1%)	15.1 KiB	✅
Claude	Thread snapshot wire	6.9 KiB	6.9 KiB	+5 B (+0.1%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.4 KiB	6.4 KiB	+15 B (+0.2%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	56.4 KiB	56.4 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	10	10	0 (0.0%)	21	✅

Baseline: 355fbd9 · PR result: 834f807 · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 109.4 KiB
• Claude decoded thread snapshot: 110.1 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.

[macroscopeapp]

Approvability

Verdict: Approved at 834f807

Macroscope's review found this PR approvable — This is a narrowly scoped one-file fix that stabilizes existing ChatMarkdown renderer identity without introducing new capabilities, schema changes, defaults, or sensitive behavior. Renderer updates still occur when the environment or thread ID changes.

^{You can add or adjust custom eligibility rules. Learn more.}

[juliusmarminge]

Problem

Follow-up to #9306. ChatMarkdown's markdownComponents memo depends on threadRef, and react-markdown uses those renderer functions as React element types. Any caller that passes a fresh-but-equal threadRef object therefore remounts every rendered <p>/<pre>/<code> in the message — which is what was destroying native text selections and dropping the cite toolbar on working threads. #9306 fixed the one caller doing that, but the component stayed one careless parseScopedThreadKey(...) away from regressing.

Fix

Memoize threadRef on environmentId + threadId at the top of ChatMarkdown, before anything downstream depends on it. Every existing use (markdownComponents, the PR-link callbacks, the media effect, the child memo comparators) now sees a stable object as long as the ids are unchanged, regardless of what the caller passes.

Verified with vp lint, apps/web typecheck, and the ChatMarkdown tests (69 passing).

---

Claude Fable 5 via Claude Code

🤖 Generated with Claude Code

---

Note

Low Risk
Small, localized React memoization in chat markdown rendering with no auth, data, or API changes.

Overview
Stabilizes threadRef inside ChatMarkdown so react-markdown does not treat every new object reference as a renderer change and remount the whole message tree (which was clearing native text selection and breaking the cite toolbar).

The prop is renamed to threadRefProp and wrapped in useMemo keyed on environmentId and threadId. Downstream logic (markdownComponents, PR-link handlers, media preview, effects) keeps the same API but sees a stable reference while the scoped thread is unchanged, even if parents pass a freshly allocated ScopedThreadRef each render.

^{Reviewed by Cursor Bugbot for commit 834f807. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Fix ChatMarkdown to key renderers on thread IDs, not object identity

Memoizes the thread reference inside ChatMarkdown using environment and thread IDs so an equivalent-but-new thread object does not create a new renderer identity. This preserves native text selections inside rendered markdown nodes across re-renders when the IDs are unchanged.

^{Macroscope summarized 834f807.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.5 KiB	13.2 KiB	−219 B (−1.6%)	15.1 KiB	✅
Codex	Thread snapshot wire	6.9 KiB	6.9 KiB	+5 B (+0.1%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.6 KiB	6.3 KiB	−224 B (−3.3%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	57.0 KiB	55.5 KiB	−1.5 KiB (−2.6%)	66.4 KiB	✅
Codex	Live turn messages	10	8	−2 (−20.0%)	21	✅
Claude	Total thread wire	13.3 KiB	13.3 KiB	+20 B (+0.1%)	15.1 KiB	✅
Claude	Thread snapshot wire	6.9 KiB	6.9 KiB	+5 B (+0.1%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.4 KiB	6.4 KiB	+15 B (+0.2%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	56.4 KiB	56.4 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	10	10	0 (0.0%)	21	✅

Baseline: 355fbd9 · PR result: 834f807 · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 109.4 KiB
• Claude decoded thread snapshot: 110.1 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.

[macroscopeapp]

Approvability

Verdict: Approved at 834f807

Macroscope's review found this PR approvable — This is a narrowly scoped one-file fix that stabilizes existing ChatMarkdown renderer identity without introducing new capabilities, schema changes, defaults, or sensitive behavior. Renderer updates still occur when the environment or thread ID changes.

^{You can add or adjust custom eligibility rules. Learn more.}

[juliusmarminge]

Problem

Follow-up to #9306. ChatMarkdown's markdownComponents memo depends on threadRef, and react-markdown uses those renderer functions as React element types. Any caller that passes a fresh-but-equal threadRef object therefore remounts every rendered <p>/<pre>/<code> in the message — which is what was destroying native text selections and dropping the cite toolbar on working threads. #9306 fixed the one caller doing that, but the component stayed one careless parseScopedThreadKey(...) away from regressing.

Fix

Memoize threadRef on environmentId + threadId at the top of ChatMarkdown, before anything downstream depends on it. Every existing use (markdownComponents, the PR-link callbacks, the media effect, the child memo comparators) now sees a stable object as long as the ids are unchanged, regardless of what the caller passes.

Verified with vp lint, apps/web typecheck, and the ChatMarkdown tests (69 passing).

---

Claude Fable 5 via Claude Code

🤖 Generated with Claude Code

---

Note

Low Risk
Small, localized React memoization in chat markdown rendering with no auth, data, or API changes.

Overview
Stabilizes threadRef inside ChatMarkdown so react-markdown does not treat every new object reference as a renderer change and remount the whole message tree (which was clearing native text selection and breaking the cite toolbar).

The prop is renamed to threadRefProp and wrapped in useMemo keyed on environmentId and threadId. Downstream logic (markdownComponents, PR-link handlers, media preview, effects) keeps the same API but sees a stable reference while the scoped thread is unchanged, even if parents pass a freshly allocated ScopedThreadRef each render.

^{Reviewed by Cursor Bugbot for commit 834f807. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Fix ChatMarkdown to key renderers on thread IDs, not object identity

Memoizes the thread reference inside ChatMarkdown using environment and thread IDs so an equivalent-but-new thread object does not create a new renderer identity. This preserves native text selections inside rendered markdown nodes across re-renders when the IDs are unchanged.

^{Macroscope summarized 834f807.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.5 KiB	13.2 KiB	−219 B (−1.6%)	15.1 KiB	✅
Codex	Thread snapshot wire	6.9 KiB	6.9 KiB	+5 B (+0.1%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.6 KiB	6.3 KiB	−224 B (−3.3%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	57.0 KiB	55.5 KiB	−1.5 KiB (−2.6%)	66.4 KiB	✅
Codex	Live turn messages	10	8	−2 (−20.0%)	21	✅
Claude	Total thread wire	13.3 KiB	13.3 KiB	+20 B (+0.1%)	15.1 KiB	✅
Claude	Thread snapshot wire	6.9 KiB	6.9 KiB	+5 B (+0.1%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.4 KiB	6.4 KiB	+15 B (+0.2%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	56.4 KiB	56.4 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	10	10	0 (0.0%)	21	✅

Baseline: 355fbd9 · PR result: 834f807 · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 109.4 KiB
• Claude decoded thread snapshot: 110.1 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.

[macroscopeapp]

Approvability

Verdict: Approved at 834f807

Macroscope's review found this PR approvable — This is a narrowly scoped one-file fix that stabilizes existing ChatMarkdown renderer identity without introducing new capabilities, schema changes, defaults, or sensitive behavior. Renderer updates still occur when the environment or thread ID changes.

^{You can add or adjust custom eligibility rules. Learn more.}

[juliusmarminge]

Problem

Follow-up to #9306. ChatMarkdown's markdownComponents memo depends on threadRef, and react-markdown uses those renderer functions as React element types. Any caller that passes a fresh-but-equal threadRef object therefore remounts every rendered <p>/<pre>/<code> in the message — which is what was destroying native text selections and dropping the cite toolbar on working threads. #9306 fixed the one caller doing that, but the component stayed one careless parseScopedThreadKey(...) away from regressing.

Fix

Memoize threadRef on environmentId + threadId at the top of ChatMarkdown, before anything downstream depends on it. Every existing use (markdownComponents, the PR-link callbacks, the media effect, the child memo comparators) now sees a stable object as long as the ids are unchanged, regardless of what the caller passes.

Verified with vp lint, apps/web typecheck, and the ChatMarkdown tests (69 passing).

---

Claude Fable 5 via Claude Code

🤖 Generated with Claude Code

---

Note

Low Risk
Small, localized React memoization in chat markdown rendering with no auth, data, or API changes.

Overview
Stabilizes threadRef inside ChatMarkdown so react-markdown does not treat every new object reference as a renderer change and remount the whole message tree (which was clearing native text selection and breaking the cite toolbar).

The prop is renamed to threadRefProp and wrapped in useMemo keyed on environmentId and threadId. Downstream logic (markdownComponents, PR-link handlers, media preview, effects) keeps the same API but sees a stable reference while the scoped thread is unchanged, even if parents pass a freshly allocated ScopedThreadRef each render.

^{Reviewed by Cursor Bugbot for commit 834f807. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Fix ChatMarkdown to key renderers on thread IDs, not object identity

Memoizes the thread reference inside ChatMarkdown using environment and thread IDs so an equivalent-but-new thread object does not create a new renderer identity. This preserves native text selections inside rendered markdown nodes across re-renders when the IDs are unchanged.

^{Macroscope summarized 834f807.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.5 KiB	13.2 KiB	−219 B (−1.6%)	15.1 KiB	✅
Codex	Thread snapshot wire	6.9 KiB	6.9 KiB	+5 B (+0.1%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.6 KiB	6.3 KiB	−224 B (−3.3%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	57.0 KiB	55.5 KiB	−1.5 KiB (−2.6%)	66.4 KiB	✅
Codex	Live turn messages	10	8	−2 (−20.0%)	21	✅
Claude	Total thread wire	13.3 KiB	13.3 KiB	+20 B (+0.1%)	15.1 KiB	✅
Claude	Thread snapshot wire	6.9 KiB	6.9 KiB	+5 B (+0.1%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.4 KiB	6.4 KiB	+15 B (+0.2%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	56.4 KiB	56.4 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	10	10	0 (0.0%)	21	✅

Baseline: 355fbd9 · PR result: 834f807 · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 109.4 KiB
• Claude decoded thread snapshot: 110.1 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.

[macroscopeapp]

Approvability

Verdict: Approved at 834f807

Macroscope's review found this PR approvable — This is a narrowly scoped one-file fix that stabilizes existing ChatMarkdown renderer identity without introducing new capabilities, schema changes, defaults, or sensitive behavior. Renderer updates still occur when the environment or thread ID changes.

^{You can add or adjust custom eligibility rules. Learn more.}

[juliusmarminge]

Problem

Follow-up to #9306. ChatMarkdown's markdownComponents memo depends on threadRef, and react-markdown uses those renderer functions as React element types. Any caller that passes a fresh-but-equal threadRef object therefore remounts every rendered <p>/<pre>/<code> in the message — which is what was destroying native text selections and dropping the cite toolbar on working threads. #9306 fixed the one caller doing that, but the component stayed one careless parseScopedThreadKey(...) away from regressing.

Fix

Memoize threadRef on environmentId + threadId at the top of ChatMarkdown, before anything downstream depends on it. Every existing use (markdownComponents, the PR-link callbacks, the media effect, the child memo comparators) now sees a stable object as long as the ids are unchanged, regardless of what the caller passes.

Verified with vp lint, apps/web typecheck, and the ChatMarkdown tests (69 passing).

---

Claude Fable 5 via Claude Code

🤖 Generated with Claude Code

---

Note

Low Risk
Small, localized React memoization in chat markdown rendering with no auth, data, or API changes.

Overview
Stabilizes threadRef inside ChatMarkdown so react-markdown does not treat every new object reference as a renderer change and remount the whole message tree (which was clearing native text selection and breaking the cite toolbar).

The prop is renamed to threadRefProp and wrapped in useMemo keyed on environmentId and threadId. Downstream logic (markdownComponents, PR-link handlers, media preview, effects) keeps the same API but sees a stable reference while the scoped thread is unchanged, even if parents pass a freshly allocated ScopedThreadRef each render.

^{Reviewed by Cursor Bugbot for commit 834f807. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Fix ChatMarkdown to key renderers on thread IDs, not object identity

Memoizes the thread reference inside ChatMarkdown using environment and thread IDs so an equivalent-but-new thread object does not create a new renderer identity. This preserves native text selections inside rendered markdown nodes across re-renders when the IDs are unchanged.

^{Macroscope summarized 834f807.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.5 KiB	13.2 KiB	−219 B (−1.6%)	15.1 KiB	✅
Codex	Thread snapshot wire	6.9 KiB	6.9 KiB	+5 B (+0.1%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.6 KiB	6.3 KiB	−224 B (−3.3%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	57.0 KiB	55.5 KiB	−1.5 KiB (−2.6%)	66.4 KiB	✅
Codex	Live turn messages	10	8	−2 (−20.0%)	21	✅
Claude	Total thread wire	13.3 KiB	13.3 KiB	+20 B (+0.1%)	15.1 KiB	✅
Claude	Thread snapshot wire	6.9 KiB	6.9 KiB	+5 B (+0.1%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.4 KiB	6.4 KiB	+15 B (+0.2%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	56.4 KiB	56.4 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	10	10	0 (0.0%)	21	✅

Baseline: 355fbd9 · PR result: 834f807 · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 109.4 KiB
• Claude decoded thread snapshot: 110.1 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.

[macroscopeapp]

Approvability

Verdict: Approved at 834f807

Macroscope's review found this PR approvable — This is a narrowly scoped one-file fix that stabilizes existing ChatMarkdown renderer identity without introducing new capabilities, schema changes, defaults, or sensitive behavior. Renderer updates still occur when the environment or thread ID changes.

^{You can add or adjust custom eligibility rules. Learn more.}

[juliusmarminge]

Problem

Follow-up to #9306. ChatMarkdown's markdownComponents memo depends on threadRef, and react-markdown uses those renderer functions as React element types. Any caller that passes a fresh-but-equal threadRef object therefore remounts every rendered <p>/<pre>/<code> in the message — which is what was destroying native text selections and dropping the cite toolbar on working threads. #9306 fixed the one caller doing that, but the component stayed one careless parseScopedThreadKey(...) away from regressing.

Fix

Memoize threadRef on environmentId + threadId at the top of ChatMarkdown, before anything downstream depends on it. Every existing use (markdownComponents, the PR-link callbacks, the media effect, the child memo comparators) now sees a stable object as long as the ids are unchanged, regardless of what the caller passes.

Verified with vp lint, apps/web typecheck, and the ChatMarkdown tests (69 passing).

---

Claude Fable 5 via Claude Code

🤖 Generated with Claude Code

---

Note

Low Risk
Small, localized React memoization in chat markdown rendering with no auth, data, or API changes.

Overview
Stabilizes threadRef inside ChatMarkdown so react-markdown does not treat every new object reference as a renderer change and remount the whole message tree (which was clearing native text selection and breaking the cite toolbar).

The prop is renamed to threadRefProp and wrapped in useMemo keyed on environmentId and threadId. Downstream logic (markdownComponents, PR-link handlers, media preview, effects) keeps the same API but sees a stable reference while the scoped thread is unchanged, even if parents pass a freshly allocated ScopedThreadRef each render.

^{Reviewed by Cursor Bugbot for commit 834f807. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Fix ChatMarkdown to key renderers on thread IDs, not object identity

Memoizes the thread reference inside ChatMarkdown using environment and thread IDs so an equivalent-but-new thread object does not create a new renderer identity. This preserves native text selections inside rendered markdown nodes across re-renders when the IDs are unchanged.

^{Macroscope summarized 834f807.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.5 KiB	13.2 KiB	−219 B (−1.6%)	15.1 KiB	✅
Codex	Thread snapshot wire	6.9 KiB	6.9 KiB	+5 B (+0.1%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.6 KiB	6.3 KiB	−224 B (−3.3%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	57.0 KiB	55.5 KiB	−1.5 KiB (−2.6%)	66.4 KiB	✅
Codex	Live turn messages	10	8	−2 (−20.0%)	21	✅
Claude	Total thread wire	13.3 KiB	13.3 KiB	+20 B (+0.1%)	15.1 KiB	✅
Claude	Thread snapshot wire	6.9 KiB	6.9 KiB	+5 B (+0.1%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.4 KiB	6.4 KiB	+15 B (+0.2%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	56.4 KiB	56.4 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	10	10	0 (0.0%)	21	✅

Baseline: 355fbd9 · PR result: 834f807 · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 109.4 KiB
• Claude decoded thread snapshot: 110.1 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.

[macroscopeapp]

Approvability

Verdict: Approved at 834f807

Macroscope's review found this PR approvable — This is a narrowly scoped one-file fix that stabilizes existing ChatMarkdown renderer identity without introducing new capabilities, schema changes, defaults, or sensitive behavior. Renderer updates still occur when the environment or thread ID changes.

^{You can add or adjust custom eligibility rules. Learn more.}