fix(server): generate durable thread titles

[t3dotgg]

Thread titles often captured the requested artifact or workflow instead of the lasting subject, making conversations harder to find later.

This switches initial and regenerated thread titles to the evaluated durable-intent prompt. It also makes the default text generation selection explicitly use GPT-5.6 Luna with low reasoning effort, matching the tested configuration.

Testing

• Focused text generation and settings tests: 85 passed
• Targeted typecheck attempted; current main is blocked by existing Effect beta.103 errors in untouched files

Built with GPT-5.6 Sol at high reasoning effort through the T3 Code Codex harness.

---

Note

Low Risk
Prompt-only title behavior change plus settings defaults; no auth or data-path changes, with coverage in prompt and settings tests.

Overview
Thread title generation is reworked so titles emphasize durable subject and outcome instead of restating the user message or naming artifacts (mocks, plans, PRs).

buildThreadTitlePrompt now uses a subject/outcome/incidental-instructions framing, editorial rules (length, reviews/research, regeneration across full thread), and optional guidance / rulesLabel on the shared buildPromptFromMessage helper. Tests and the Claude integration test expectations were updated for the new prompt text.

Default text generation explicitly defaults to GPT-5.6 Luna with reasoningEffort: low: DEFAULT_TEXT_GENERATION_REASONING_EFFORT is exported from contracts, wired into ServerSettings.textGenerationModelSelection, and used as the Codex CLI fallback when the selection omits effort.

^{Reviewed by Cursor Bugbot for commit ed1863e. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Rewrite thread title prompts to produce durable, subject-focused titles

• Rewrites buildThreadTitlePrompt in TextGenerationPrompts.ts with a new guidance section and editorial rules that direct the model to extract subject/outcome rather than summarizing the user's initial request verbatim.
• Extends buildPromptFromMessage to accept optional guidance lines and a custom rulesLabel, used to inject the new editorial section.
• Exports DEFAULT_TEXT_GENERATION_REASONING_EFFORT ('low') from contracts and uses it as the default reasoningEffort in both CodexTextGeneration and the default textGenerationModelSelection in ServerSettings.
• Behavioral Change: ServerSettings decoded without an explicit textGenerationModelSelection now includes a reasoningEffort: 'low' option by default.

^{Macroscope summarized ed1863e.}

[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: 69286312-7450-40e8-afcd-ee904a106a50

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.}

[macroscopeapp]

Approvability

Verdict: Approved ed1863e

This PR improves prompts for thread title generation and consolidates a reasoning effort constant to the shared contracts package. The changes affect AI-generated title quality without modifying application logic, and the constant consolidation maintains the existing 'low' reasoning effort value.

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

Dismissing prior approval to re-evaluate ed1863e