feat: remove Ultrathink

[t3dotgg]

Ultrathink is dead, but T3 Code still offered it as a Claude reasoning mode and treated the word as a prompt-controlled feature.

Remove the option and its prompt rewriting, composer styling, client controls, contract metadata, and adapter handling across web, desktop, mobile, and server. Existing saved values now fall back to the model default.

Built with GPT-5.6 Sol in the Codex harness through T3 Code.

---

Note

Medium Risk
Touches how Claude turns are built and how effort options are modeled; behavior change for anyone who relied on ultrathink or saved ultrathink selections (now invalid catalog options, no automatic prefix).

Overview
Removes Ultrathink as a Claude reasoning option and deletes the whole prompt-injected effort path (promptInjectedValues, Ultrathink: prefixing, composer chrome).

Server & contracts: Ultrathink is dropped from Claude model catalogs; promptInjectedValues is removed from select option descriptors in contracts and buildSelectOptionDescriptor. ClaudeAdapter sends trimmed user text only (no effort→prefix mapping); CLI effort normalization no longer special-cases ultrathink. Related adapter tests are removed.

Web: Outgoing sends use trimmed prompt text instead of formatOutgoingPrompt. The traits picker no longer reads or edits the composer for ultrathink; getSendContext drops selectedPromptEffort; ultrathink CSS/animations are deleted.

Mobile: Thread settings filters selectable effort options via HIDDEN_EFFORT_OPTION_IDS only (e.g. ultracode), not prompt-injected values.

Shared:resolvePromptInjectedEffort, applyClaudePromptEffortPrefix, and isClaudeUltrathinkPrompt are removed; invalid effort values no longer get silently remapped via prompt-injection metadata.

^{Reviewed by Cursor Bugbot for commit 866476e. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Remove Ultrathink effort option from Claude models

• Removes the ultrathink effort option from all Claude model descriptors in ClaudeProvider.ts and the shared model contract.
• Strips prompt-prefix injection logic (applyClaudePromptEffortPrefix, resolvePromptInjectedEffort, isClaudeUltrathinkPrompt) from the server, web, and shared packages; outgoing prompts are now sent as trimmed user input only.
• Removes promptInjectedValues from the SelectProviderOptionDescriptor contract schema and all related runtime logic.
• Removes Ultrathink-specific CSS animations and styling from index.css.
• Behavioral Change: Selecting a high-effort mode no longer injects an Ultrathink: prefix into the prompt text; the traits UI no longer modifies composer content.

^{Macroscope summarized 866476e.}

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 829d03e0-1fec-4d1d-b97d-60f8f6e8d74f

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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

[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	11.3 KiB	11.3 KiB	+18 B (+0.2%)	15.1 KiB	✅
Codex	Thread snapshot wire	5.5 KiB	5.5 KiB	+7 B (+0.1%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	5.9 KiB	5.9 KiB	+11 B (+0.2%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	49.7 KiB	49.7 KiB	0 B (0.0%)	66.4 KiB	✅
Codex	Live turn messages	16	16	0 (0.0%)	21	✅
Claude	Total thread wire	11.3 KiB	11.3 KiB	+17 B (+0.1%)	15.1 KiB	✅
Claude	Thread snapshot wire	5.5 KiB	5.5 KiB	−2 B (−0.0%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	5.8 KiB	5.9 KiB	+19 B (+0.3%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	50.6 KiB	50.6 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	16	16	0 (0.0%)	21	✅

Baseline: db1507e · PR result: 866476e · 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: 94.6 KiB
• Claude decoded thread snapshot: 95.4 KiB

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

[macroscopeapp]

Approvability

Verdict: Needs human review

This PR removes the Ultrathink feature entirely, including UI options, prompt transformation logic, and visual styling. While the changes are mechanically straightforward deletions, removing a user-facing feature constitutes a significant runtime behavior change that warrants human confirmation.

^{You can customize Macroscope's approvability policy. Learn more.}

[t3dotgg]

Ultrathink is dead, but T3 Code still offered it as a Claude reasoning mode and treated the word as a prompt-controlled feature.

Remove the option and its prompt rewriting, composer styling, client controls, contract metadata, and adapter handling across web, desktop, mobile, and server. Existing saved values now fall back to the model default.

Built with GPT-5.6 Sol in the Codex harness through T3 Code.

---

Note

Medium Risk
Touches how Claude turns are built and how effort options are modeled; behavior change for anyone who relied on ultrathink or saved ultrathink selections (now invalid catalog options, no automatic prefix).

Overview
Removes Ultrathink as a Claude reasoning option and deletes the whole prompt-injected effort path (promptInjectedValues, Ultrathink: prefixing, composer chrome).

Server & contracts: Ultrathink is dropped from Claude model catalogs; promptInjectedValues is removed from select option descriptors in contracts and buildSelectOptionDescriptor. ClaudeAdapter sends trimmed user text only (no effort→prefix mapping); CLI effort normalization no longer special-cases ultrathink. Related adapter tests are removed.

Web: Outgoing sends use trimmed prompt text instead of formatOutgoingPrompt. The traits picker no longer reads or edits the composer for ultrathink; getSendContext drops selectedPromptEffort; ultrathink CSS/animations are deleted.

Mobile: Thread settings filters selectable effort options via HIDDEN_EFFORT_OPTION_IDS only (e.g. ultracode), not prompt-injected values.

Shared:resolvePromptInjectedEffort, applyClaudePromptEffortPrefix, and isClaudeUltrathinkPrompt are removed; invalid effort values no longer get silently remapped via prompt-injection metadata.

^{Reviewed by Cursor Bugbot for commit 866476e. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Remove Ultrathink effort option from Claude models

• Removes the ultrathink effort option from all Claude model descriptors in ClaudeProvider.ts and the shared model contract.
• Strips prompt-prefix injection logic (applyClaudePromptEffortPrefix, resolvePromptInjectedEffort, isClaudeUltrathinkPrompt) from the server, web, and shared packages; outgoing prompts are now sent as trimmed user input only.
• Removes promptInjectedValues from the SelectProviderOptionDescriptor contract schema and all related runtime logic.
• Removes Ultrathink-specific CSS animations and styling from index.css.
• Behavioral Change: Selecting a high-effort mode no longer injects an Ultrathink: prefix into the prompt text; the traits UI no longer modifies composer content.

^{Macroscope summarized 866476e.}

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 829d03e0-1fec-4d1d-b97d-60f8f6e8d74f

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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

[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	11.3 KiB	11.3 KiB	+18 B (+0.2%)	15.1 KiB	✅
Codex	Thread snapshot wire	5.5 KiB	5.5 KiB	+7 B (+0.1%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	5.9 KiB	5.9 KiB	+11 B (+0.2%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	49.7 KiB	49.7 KiB	0 B (0.0%)	66.4 KiB	✅
Codex	Live turn messages	16	16	0 (0.0%)	21	✅
Claude	Total thread wire	11.3 KiB	11.3 KiB	+17 B (+0.1%)	15.1 KiB	✅
Claude	Thread snapshot wire	5.5 KiB	5.5 KiB	−2 B (−0.0%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	5.8 KiB	5.9 KiB	+19 B (+0.3%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	50.6 KiB	50.6 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	16	16	0 (0.0%)	21	✅

Baseline: db1507e · PR result: 866476e · 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: 94.6 KiB
• Claude decoded thread snapshot: 95.4 KiB

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

[macroscopeapp]

Approvability

Verdict: Needs human review

This PR removes the Ultrathink feature entirely, including UI options, prompt transformation logic, and visual styling. While the changes are mechanically straightforward deletions, removing a user-facing feature constitutes a significant runtime behavior change that warrants human confirmation.

^{You can customize Macroscope's approvability policy. Learn more.}

[t3dotgg]

Ultrathink is dead, but T3 Code still offered it as a Claude reasoning mode and treated the word as a prompt-controlled feature.

Remove the option and its prompt rewriting, composer styling, client controls, contract metadata, and adapter handling across web, desktop, mobile, and server. Existing saved values now fall back to the model default.

Built with GPT-5.6 Sol in the Codex harness through T3 Code.

---

Note

Medium Risk
Touches how Claude turns are built and how effort options are modeled; behavior change for anyone who relied on ultrathink or saved ultrathink selections (now invalid catalog options, no automatic prefix).

Overview
Removes Ultrathink as a Claude reasoning option and deletes the whole prompt-injected effort path (promptInjectedValues, Ultrathink: prefixing, composer chrome).

Server & contracts: Ultrathink is dropped from Claude model catalogs; promptInjectedValues is removed from select option descriptors in contracts and buildSelectOptionDescriptor. ClaudeAdapter sends trimmed user text only (no effort→prefix mapping); CLI effort normalization no longer special-cases ultrathink. Related adapter tests are removed.

Web: Outgoing sends use trimmed prompt text instead of formatOutgoingPrompt. The traits picker no longer reads or edits the composer for ultrathink; getSendContext drops selectedPromptEffort; ultrathink CSS/animations are deleted.

Mobile: Thread settings filters selectable effort options via HIDDEN_EFFORT_OPTION_IDS only (e.g. ultracode), not prompt-injected values.

Shared:resolvePromptInjectedEffort, applyClaudePromptEffortPrefix, and isClaudeUltrathinkPrompt are removed; invalid effort values no longer get silently remapped via prompt-injection metadata.

^{Reviewed by Cursor Bugbot for commit 866476e. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Remove Ultrathink effort option from Claude models

• Removes the ultrathink effort option from all Claude model descriptors in ClaudeProvider.ts and the shared model contract.
• Strips prompt-prefix injection logic (applyClaudePromptEffortPrefix, resolvePromptInjectedEffort, isClaudeUltrathinkPrompt) from the server, web, and shared packages; outgoing prompts are now sent as trimmed user input only.
• Removes promptInjectedValues from the SelectProviderOptionDescriptor contract schema and all related runtime logic.
• Removes Ultrathink-specific CSS animations and styling from index.css.
• Behavioral Change: Selecting a high-effort mode no longer injects an Ultrathink: prefix into the prompt text; the traits UI no longer modifies composer content.

^{Macroscope summarized 866476e.}

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 829d03e0-1fec-4d1d-b97d-60f8f6e8d74f

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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

[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	11.3 KiB	11.3 KiB	+18 B (+0.2%)	15.1 KiB	✅
Codex	Thread snapshot wire	5.5 KiB	5.5 KiB	+7 B (+0.1%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	5.9 KiB	5.9 KiB	+11 B (+0.2%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	49.7 KiB	49.7 KiB	0 B (0.0%)	66.4 KiB	✅
Codex	Live turn messages	16	16	0 (0.0%)	21	✅
Claude	Total thread wire	11.3 KiB	11.3 KiB	+17 B (+0.1%)	15.1 KiB	✅
Claude	Thread snapshot wire	5.5 KiB	5.5 KiB	−2 B (−0.0%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	5.8 KiB	5.9 KiB	+19 B (+0.3%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	50.6 KiB	50.6 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	16	16	0 (0.0%)	21	✅

Baseline: db1507e · PR result: 866476e · 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: 94.6 KiB
• Claude decoded thread snapshot: 95.4 KiB

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

[macroscopeapp]

Approvability

Verdict: Needs human review

This PR removes the Ultrathink feature entirely, including UI options, prompt transformation logic, and visual styling. While the changes are mechanically straightforward deletions, removing a user-facing feature constitutes a significant runtime behavior change that warrants human confirmation.

^{You can customize Macroscope's approvability policy. Learn more.}

[t3dotgg]

Ultrathink is dead, but T3 Code still offered it as a Claude reasoning mode and treated the word as a prompt-controlled feature.

Remove the option and its prompt rewriting, composer styling, client controls, contract metadata, and adapter handling across web, desktop, mobile, and server. Existing saved values now fall back to the model default.

Built with GPT-5.6 Sol in the Codex harness through T3 Code.

---

Note

Medium Risk
Touches how Claude turns are built and how effort options are modeled; behavior change for anyone who relied on ultrathink or saved ultrathink selections (now invalid catalog options, no automatic prefix).

Overview
Removes Ultrathink as a Claude reasoning option and deletes the whole prompt-injected effort path (promptInjectedValues, Ultrathink: prefixing, composer chrome).

Server & contracts: Ultrathink is dropped from Claude model catalogs; promptInjectedValues is removed from select option descriptors in contracts and buildSelectOptionDescriptor. ClaudeAdapter sends trimmed user text only (no effort→prefix mapping); CLI effort normalization no longer special-cases ultrathink. Related adapter tests are removed.

Web: Outgoing sends use trimmed prompt text instead of formatOutgoingPrompt. The traits picker no longer reads or edits the composer for ultrathink; getSendContext drops selectedPromptEffort; ultrathink CSS/animations are deleted.

Mobile: Thread settings filters selectable effort options via HIDDEN_EFFORT_OPTION_IDS only (e.g. ultracode), not prompt-injected values.

Shared:resolvePromptInjectedEffort, applyClaudePromptEffortPrefix, and isClaudeUltrathinkPrompt are removed; invalid effort values no longer get silently remapped via prompt-injection metadata.

^{Reviewed by Cursor Bugbot for commit 866476e. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Remove Ultrathink effort option from Claude models

• Removes the ultrathink effort option from all Claude model descriptors in ClaudeProvider.ts and the shared model contract.
• Strips prompt-prefix injection logic (applyClaudePromptEffortPrefix, resolvePromptInjectedEffort, isClaudeUltrathinkPrompt) from the server, web, and shared packages; outgoing prompts are now sent as trimmed user input only.
• Removes promptInjectedValues from the SelectProviderOptionDescriptor contract schema and all related runtime logic.
• Removes Ultrathink-specific CSS animations and styling from index.css.
• Behavioral Change: Selecting a high-effort mode no longer injects an Ultrathink: prefix into the prompt text; the traits UI no longer modifies composer content.

^{Macroscope summarized 866476e.}

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 829d03e0-1fec-4d1d-b97d-60f8f6e8d74f

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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

[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	11.3 KiB	11.3 KiB	+18 B (+0.2%)	15.1 KiB	✅
Codex	Thread snapshot wire	5.5 KiB	5.5 KiB	+7 B (+0.1%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	5.9 KiB	5.9 KiB	+11 B (+0.2%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	49.7 KiB	49.7 KiB	0 B (0.0%)	66.4 KiB	✅
Codex	Live turn messages	16	16	0 (0.0%)	21	✅
Claude	Total thread wire	11.3 KiB	11.3 KiB	+17 B (+0.1%)	15.1 KiB	✅
Claude	Thread snapshot wire	5.5 KiB	5.5 KiB	−2 B (−0.0%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	5.8 KiB	5.9 KiB	+19 B (+0.3%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	50.6 KiB	50.6 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	16	16	0 (0.0%)	21	✅

Baseline: db1507e · PR result: 866476e · 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: 94.6 KiB
• Claude decoded thread snapshot: 95.4 KiB

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

[macroscopeapp]

Approvability

Verdict: Needs human review

This PR removes the Ultrathink feature entirely, including UI options, prompt transformation logic, and visual styling. While the changes are mechanically straightforward deletions, removing a user-facing feature constitutes a significant runtime behavior change that warrants human confirmation.

^{You can customize Macroscope's approvability policy. Learn more.}

[t3dotgg]

Ultrathink is dead, but T3 Code still offered it as a Claude reasoning mode and treated the word as a prompt-controlled feature.

Remove the option and its prompt rewriting, composer styling, client controls, contract metadata, and adapter handling across web, desktop, mobile, and server. Existing saved values now fall back to the model default.

Built with GPT-5.6 Sol in the Codex harness through T3 Code.

---

Note

Medium Risk
Touches how Claude turns are built and how effort options are modeled; behavior change for anyone who relied on ultrathink or saved ultrathink selections (now invalid catalog options, no automatic prefix).

Overview
Removes Ultrathink as a Claude reasoning option and deletes the whole prompt-injected effort path (promptInjectedValues, Ultrathink: prefixing, composer chrome).

Server & contracts: Ultrathink is dropped from Claude model catalogs; promptInjectedValues is removed from select option descriptors in contracts and buildSelectOptionDescriptor. ClaudeAdapter sends trimmed user text only (no effort→prefix mapping); CLI effort normalization no longer special-cases ultrathink. Related adapter tests are removed.

Web: Outgoing sends use trimmed prompt text instead of formatOutgoingPrompt. The traits picker no longer reads or edits the composer for ultrathink; getSendContext drops selectedPromptEffort; ultrathink CSS/animations are deleted.

Mobile: Thread settings filters selectable effort options via HIDDEN_EFFORT_OPTION_IDS only (e.g. ultracode), not prompt-injected values.

Shared:resolvePromptInjectedEffort, applyClaudePromptEffortPrefix, and isClaudeUltrathinkPrompt are removed; invalid effort values no longer get silently remapped via prompt-injection metadata.

^{Reviewed by Cursor Bugbot for commit 866476e. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Remove Ultrathink effort option from Claude models

• Removes the ultrathink effort option from all Claude model descriptors in ClaudeProvider.ts and the shared model contract.
• Strips prompt-prefix injection logic (applyClaudePromptEffortPrefix, resolvePromptInjectedEffort, isClaudeUltrathinkPrompt) from the server, web, and shared packages; outgoing prompts are now sent as trimmed user input only.
• Removes promptInjectedValues from the SelectProviderOptionDescriptor contract schema and all related runtime logic.
• Removes Ultrathink-specific CSS animations and styling from index.css.
• Behavioral Change: Selecting a high-effort mode no longer injects an Ultrathink: prefix into the prompt text; the traits UI no longer modifies composer content.

^{Macroscope summarized 866476e.}

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 829d03e0-1fec-4d1d-b97d-60f8f6e8d74f

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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

[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	11.3 KiB	11.3 KiB	+18 B (+0.2%)	15.1 KiB	✅
Codex	Thread snapshot wire	5.5 KiB	5.5 KiB	+7 B (+0.1%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	5.9 KiB	5.9 KiB	+11 B (+0.2%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	49.7 KiB	49.7 KiB	0 B (0.0%)	66.4 KiB	✅
Codex	Live turn messages	16	16	0 (0.0%)	21	✅
Claude	Total thread wire	11.3 KiB	11.3 KiB	+17 B (+0.1%)	15.1 KiB	✅
Claude	Thread snapshot wire	5.5 KiB	5.5 KiB	−2 B (−0.0%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	5.8 KiB	5.9 KiB	+19 B (+0.3%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	50.6 KiB	50.6 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	16	16	0 (0.0%)	21	✅

Baseline: db1507e · PR result: 866476e · 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: 94.6 KiB
• Claude decoded thread snapshot: 95.4 KiB

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

[macroscopeapp]

Approvability

Verdict: Needs human review

This PR removes the Ultrathink feature entirely, including UI options, prompt transformation logic, and visual styling. While the changes are mechanically straightforward deletions, removing a user-facing feature constitutes a significant runtime behavior change that warrants human confirmation.

^{You can customize Macroscope's approvability policy. Learn more.}

[t3dotgg]

Ultrathink is dead, but T3 Code still offered it as a Claude reasoning mode and treated the word as a prompt-controlled feature.

Remove the option and its prompt rewriting, composer styling, client controls, contract metadata, and adapter handling across web, desktop, mobile, and server. Existing saved values now fall back to the model default.

Built with GPT-5.6 Sol in the Codex harness through T3 Code.

---

Note

Medium Risk
Touches how Claude turns are built and how effort options are modeled; behavior change for anyone who relied on ultrathink or saved ultrathink selections (now invalid catalog options, no automatic prefix).

Overview
Removes Ultrathink as a Claude reasoning option and deletes the whole prompt-injected effort path (promptInjectedValues, Ultrathink: prefixing, composer chrome).

Server & contracts: Ultrathink is dropped from Claude model catalogs; promptInjectedValues is removed from select option descriptors in contracts and buildSelectOptionDescriptor. ClaudeAdapter sends trimmed user text only (no effort→prefix mapping); CLI effort normalization no longer special-cases ultrathink. Related adapter tests are removed.

Web: Outgoing sends use trimmed prompt text instead of formatOutgoingPrompt. The traits picker no longer reads or edits the composer for ultrathink; getSendContext drops selectedPromptEffort; ultrathink CSS/animations are deleted.

Mobile: Thread settings filters selectable effort options via HIDDEN_EFFORT_OPTION_IDS only (e.g. ultracode), not prompt-injected values.

Shared:resolvePromptInjectedEffort, applyClaudePromptEffortPrefix, and isClaudeUltrathinkPrompt are removed; invalid effort values no longer get silently remapped via prompt-injection metadata.

^{Reviewed by Cursor Bugbot for commit 866476e. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Remove Ultrathink effort option from Claude models

• Removes the ultrathink effort option from all Claude model descriptors in ClaudeProvider.ts and the shared model contract.
• Strips prompt-prefix injection logic (applyClaudePromptEffortPrefix, resolvePromptInjectedEffort, isClaudeUltrathinkPrompt) from the server, web, and shared packages; outgoing prompts are now sent as trimmed user input only.
• Removes promptInjectedValues from the SelectProviderOptionDescriptor contract schema and all related runtime logic.
• Removes Ultrathink-specific CSS animations and styling from index.css.
• Behavioral Change: Selecting a high-effort mode no longer injects an Ultrathink: prefix into the prompt text; the traits UI no longer modifies composer content.

^{Macroscope summarized 866476e.}

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 829d03e0-1fec-4d1d-b97d-60f8f6e8d74f

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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

[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	11.3 KiB	11.3 KiB	+18 B (+0.2%)	15.1 KiB	✅
Codex	Thread snapshot wire	5.5 KiB	5.5 KiB	+7 B (+0.1%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	5.9 KiB	5.9 KiB	+11 B (+0.2%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	49.7 KiB	49.7 KiB	0 B (0.0%)	66.4 KiB	✅
Codex	Live turn messages	16	16	0 (0.0%)	21	✅
Claude	Total thread wire	11.3 KiB	11.3 KiB	+17 B (+0.1%)	15.1 KiB	✅
Claude	Thread snapshot wire	5.5 KiB	5.5 KiB	−2 B (−0.0%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	5.8 KiB	5.9 KiB	+19 B (+0.3%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	50.6 KiB	50.6 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	16	16	0 (0.0%)	21	✅

Baseline: db1507e · PR result: 866476e · 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: 94.6 KiB
• Claude decoded thread snapshot: 95.4 KiB

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

[macroscopeapp]

Approvability

Verdict: Needs human review

This PR removes the Ultrathink feature entirely, including UI options, prompt transformation logic, and visual styling. While the changes are mechanically straightforward deletions, removing a user-facing feature constitutes a significant runtime behavior change that warrants human confirmation.

^{You can customize Macroscope's approvability policy. Learn more.}

[t3dotgg]

Ultrathink is dead, but T3 Code still offered it as a Claude reasoning mode and treated the word as a prompt-controlled feature.

Remove the option and its prompt rewriting, composer styling, client controls, contract metadata, and adapter handling across web, desktop, mobile, and server. Existing saved values now fall back to the model default.

Built with GPT-5.6 Sol in the Codex harness through T3 Code.

---

Note

Medium Risk
Touches how Claude turns are built and how effort options are modeled; behavior change for anyone who relied on ultrathink or saved ultrathink selections (now invalid catalog options, no automatic prefix).

Overview
Removes Ultrathink as a Claude reasoning option and deletes the whole prompt-injected effort path (promptInjectedValues, Ultrathink: prefixing, composer chrome).

Server & contracts: Ultrathink is dropped from Claude model catalogs; promptInjectedValues is removed from select option descriptors in contracts and buildSelectOptionDescriptor. ClaudeAdapter sends trimmed user text only (no effort→prefix mapping); CLI effort normalization no longer special-cases ultrathink. Related adapter tests are removed.

Web: Outgoing sends use trimmed prompt text instead of formatOutgoingPrompt. The traits picker no longer reads or edits the composer for ultrathink; getSendContext drops selectedPromptEffort; ultrathink CSS/animations are deleted.

Mobile: Thread settings filters selectable effort options via HIDDEN_EFFORT_OPTION_IDS only (e.g. ultracode), not prompt-injected values.

Shared:resolvePromptInjectedEffort, applyClaudePromptEffortPrefix, and isClaudeUltrathinkPrompt are removed; invalid effort values no longer get silently remapped via prompt-injection metadata.

^{Reviewed by Cursor Bugbot for commit 866476e. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Remove Ultrathink effort option from Claude models

• Removes the ultrathink effort option from all Claude model descriptors in ClaudeProvider.ts and the shared model contract.
• Strips prompt-prefix injection logic (applyClaudePromptEffortPrefix, resolvePromptInjectedEffort, isClaudeUltrathinkPrompt) from the server, web, and shared packages; outgoing prompts are now sent as trimmed user input only.
• Removes promptInjectedValues from the SelectProviderOptionDescriptor contract schema and all related runtime logic.
• Removes Ultrathink-specific CSS animations and styling from index.css.
• Behavioral Change: Selecting a high-effort mode no longer injects an Ultrathink: prefix into the prompt text; the traits UI no longer modifies composer content.

^{Macroscope summarized 866476e.}

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 829d03e0-1fec-4d1d-b97d-60f8f6e8d74f

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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

[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	11.3 KiB	11.3 KiB	+18 B (+0.2%)	15.1 KiB	✅
Codex	Thread snapshot wire	5.5 KiB	5.5 KiB	+7 B (+0.1%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	5.9 KiB	5.9 KiB	+11 B (+0.2%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	49.7 KiB	49.7 KiB	0 B (0.0%)	66.4 KiB	✅
Codex	Live turn messages	16	16	0 (0.0%)	21	✅
Claude	Total thread wire	11.3 KiB	11.3 KiB	+17 B (+0.1%)	15.1 KiB	✅
Claude	Thread snapshot wire	5.5 KiB	5.5 KiB	−2 B (−0.0%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	5.8 KiB	5.9 KiB	+19 B (+0.3%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	50.6 KiB	50.6 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	16	16	0 (0.0%)	21	✅

Baseline: db1507e · PR result: 866476e · 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: 94.6 KiB
• Claude decoded thread snapshot: 95.4 KiB

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

[macroscopeapp]

Approvability

Verdict: Needs human review

This PR removes the Ultrathink feature entirely, including UI options, prompt transformation logic, and visual styling. While the changes are mechanically straightforward deletions, removing a user-facing feature constitutes a significant runtime behavior change that warrants human confirmation.

^{You can customize Macroscope's approvability policy. Learn more.}

[t3dotgg]

Ultrathink is dead, but T3 Code still offered it as a Claude reasoning mode and treated the word as a prompt-controlled feature.

Remove the option and its prompt rewriting, composer styling, client controls, contract metadata, and adapter handling across web, desktop, mobile, and server. Existing saved values now fall back to the model default.

Built with GPT-5.6 Sol in the Codex harness through T3 Code.

---

Note

Medium Risk
Touches how Claude turns are built and how effort options are modeled; behavior change for anyone who relied on ultrathink or saved ultrathink selections (now invalid catalog options, no automatic prefix).

Overview
Removes Ultrathink as a Claude reasoning option and deletes the whole prompt-injected effort path (promptInjectedValues, Ultrathink: prefixing, composer chrome).

Server & contracts: Ultrathink is dropped from Claude model catalogs; promptInjectedValues is removed from select option descriptors in contracts and buildSelectOptionDescriptor. ClaudeAdapter sends trimmed user text only (no effort→prefix mapping); CLI effort normalization no longer special-cases ultrathink. Related adapter tests are removed.

Web: Outgoing sends use trimmed prompt text instead of formatOutgoingPrompt. The traits picker no longer reads or edits the composer for ultrathink; getSendContext drops selectedPromptEffort; ultrathink CSS/animations are deleted.

Mobile: Thread settings filters selectable effort options via HIDDEN_EFFORT_OPTION_IDS only (e.g. ultracode), not prompt-injected values.

Shared:resolvePromptInjectedEffort, applyClaudePromptEffortPrefix, and isClaudeUltrathinkPrompt are removed; invalid effort values no longer get silently remapped via prompt-injection metadata.

^{Reviewed by Cursor Bugbot for commit 866476e. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Remove Ultrathink effort option from Claude models

• Removes the ultrathink effort option from all Claude model descriptors in ClaudeProvider.ts and the shared model contract.
• Strips prompt-prefix injection logic (applyClaudePromptEffortPrefix, resolvePromptInjectedEffort, isClaudeUltrathinkPrompt) from the server, web, and shared packages; outgoing prompts are now sent as trimmed user input only.
• Removes promptInjectedValues from the SelectProviderOptionDescriptor contract schema and all related runtime logic.
• Removes Ultrathink-specific CSS animations and styling from index.css.
• Behavioral Change: Selecting a high-effort mode no longer injects an Ultrathink: prefix into the prompt text; the traits UI no longer modifies composer content.

^{Macroscope summarized 866476e.}

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 829d03e0-1fec-4d1d-b97d-60f8f6e8d74f

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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

[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	11.3 KiB	11.3 KiB	+18 B (+0.2%)	15.1 KiB	✅
Codex	Thread snapshot wire	5.5 KiB	5.5 KiB	+7 B (+0.1%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	5.9 KiB	5.9 KiB	+11 B (+0.2%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	49.7 KiB	49.7 KiB	0 B (0.0%)	66.4 KiB	✅
Codex	Live turn messages	16	16	0 (0.0%)	21	✅
Claude	Total thread wire	11.3 KiB	11.3 KiB	+17 B (+0.1%)	15.1 KiB	✅
Claude	Thread snapshot wire	5.5 KiB	5.5 KiB	−2 B (−0.0%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	5.8 KiB	5.9 KiB	+19 B (+0.3%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	50.6 KiB	50.6 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	16	16	0 (0.0%)	21	✅

Baseline: db1507e · PR result: 866476e · 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: 94.6 KiB
• Claude decoded thread snapshot: 95.4 KiB

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

[macroscopeapp]

Approvability

Verdict: Needs human review

This PR removes the Ultrathink feature entirely, including UI options, prompt transformation logic, and visual styling. While the changes are mechanically straightforward deletions, removing a user-facing feature constitutes a significant runtime behavior change that warrants human confirmation.

^{You can customize Macroscope's approvability policy. Learn more.}