fix(claude): separate reasoning summary parts in streamed thinking

[DevMello]

Summary

On the native ChatGPT backend the upstream Responses stream is translated straight into Anthropic SSE. Reasoning summaries arrive there in parts, each its own paragraph with a bold headline. The streaming translator folded every part into one thinking block with no separator, so thinking text rendered as run-on lines like "wants X.Planning the fix". The non-streaming translator in the same file already joins parts with a blank line. The stream now tracks part identity on the delta frames and emits the same blank-line separator at part boundaries. Frames without part indices keep a constant key and behave as before, and a new reasoning item still opens its own block.

Verification

• Two new tests in tests/claude-outbound.test.ts: multi-part summaries match the JSON path's joined text, and same-part or index-free deltas never get a separator.
• Repro harness: streamed, collected, and JSON-path thinking text are now identical for the same upstream frames.
• bun run test, typecheck, lint:gui, privacy:scan.

Checklist

• Scope stays focused and avoids unrelated cleanup.
• Docs or release notes were updated when needed.
• Security-sensitive changes were reviewed for secrets, auth, and unsafe defaults.

Summary by CodeRabbit

• Bug Fixes

  • Improved streamed reasoning display by clearly separating distinct summary sections.
  • Preserved continuous text when updates belong to the same section or lack section indicators.
  • Aligned streaming and non-streaming reasoning output for consistent results.

• Tests

  • Added coverage for multiple reasoning sections and streaming behavior.

[coderabbitai]

Warning

Review limit reached

@DevMello, you've reached your PR review limit, so we couldn't start this review.

Next review available in: 14 minutes

Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available.
You're only billed for reviews past your plan's rate limits ($0.25/file).

How can I continue?

After more reviews become available, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews.

How do review limits work?

CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability.

For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window.

Please refer docs for additional details.

Review details

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: 8e5fb602-eb44-4b49-a1da-0386ae97e924

📥 Commits

Reviewing files that changed from the base of the PR and between b23cc1f and 216a4e0.

📒 Files selected for processing (1)

• tests/claude-outbound.test.ts

📝 Walkthrough

Walkthrough

Changes

Reasoning stream handling

Layer / File(s)	Summary
Track and separate reasoning parts src/claude/outbound.ts:190-191, src/claude/outbound.ts:358-372	OpenBlock tracks the current reasoning item and part. Streaming deltas insert \n\n when the key changes, while repeated or index-free parts remain continuous.
Validate reasoning aggregation tests/claude-outbound.test.ts:210-260	Tests cover distinct summary parts, separate reasoning items, repeated indexed parts, index-free deltas, and streaming/non-streaming parity.

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

Suggested reviewers: lidge-jun, snowyukitty, ingwannu

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly and concisely describes the main change: separating reasoning summary parts in streamed Claude thinking.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

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

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@tests/claude-outbound.test.ts`:
- Around line 242-260: Add coverage in the “same-part deltas and index-free
reasoning frames never get a separator” test for indexed reasoning content by
emitting deltas with content_index 0 and 1. Collect the converted message and
assert its thinking content contains the two parts separated by "\n\n",
exercising the content_index branch while preserving the existing same-part and
index-free assertions.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: 8b37aca0-694d-480d-a5f3-7c2226c9e0ff

📥 Commits

Reviewing files that changed from the base of the PR and between 6a7351b and b23cc1f.

📒 Files selected for processing (2)

• src/claude/outbound.ts
• tests/claude-outbound.test.ts

[lidge-jun]

Carried onto the review stack as #953 (stack 3/3), unmodified.

Your commits were taken with git cherry-pick -x, so they keep your authorship — git log --format='%an' on the stack branch shows you, not me. No content was changed; the diff on the stack is byte-identical to what you wrote here, and it applied to dev with no conflict resolution.

Verified on the stack: bun x tsc --noEmit exit 0, and the full suite at 7691 pass / 8 skip / 0 fail across 507 files.

This PR stays open until #953 lands. If a maintainer prefers to take yours directly instead, that path is unaffected — the stack commits get dropped and this one merges. Once #953 merges I'll close this as carried, with the credit already in the commit history rather than in a comment.

Stack: #951 (plan, base dev) → #952 (#908 long-context pricing) → #953 (this carry). Review bottom-up.

Thanks for the fix.

[Wibias]

Leave it as draft. Thanks. It doesnt need any more changes as you can see in juns comment on which you didn't respond to.