fix: preserve thinking block signatures across three independent corruption paths

[chan1103]

Issue for this PR

Closes #13286
Related: #18078, #16748, #14393

Type of change

• Bug fix
• New feature
• Refactor / code improvement
• Documentation

What does this PR do?

Extended thinking sessions fail intermittently with "thinking blocks cannot be modified." I traced the message pipeline and found three independent places where signatures get corrupted.

1. differentModel guard strips reasoning metadata (message-v2.ts)

When the current model differs from the stored message's model, toModelMessages drops providerMetadata from all parts — including reasoning parts that carry thinking block signatures. Common with plugin-based model routing.

Fix: reasoning parts always pass their metadata through, bypassing the guard. Text/tool parts keep the guard. This is safe because provider metadata is namespaced (anthropic: { signature }) — other providers ignore unknown keys.

2. trimEnd() mutates reasoning text (processor.ts)

The reasoning-end handler trims trailing whitespace. The signature was computed against the original text, so any mutation invalidates it.

Fix: remove the trimEnd(). Reasoning text arrives final from the stream.

3. normalizeMessages removes signed empty parts (transform.ts)

The Anthropic compat filter (c285304) removes parts where text === "". This catches redacted_thinking blocks — empty text but valid signatures in providerOptions. Removing them shifts positions, and Anthropic validates positionally (#16748).

Fix: keep reasoning parts that have providerOptions. Still remove truly empty ones.

Each cause triggers independently under different conditions, which is why fixing just one doesn't eliminate the error. PR #14393 covers cause 1, PR #12131 covers cause 2 — neither covers cause 3, and this PR covers all three.

How did you verify your code works?

Added 4 tests across 3 files. Each test fails against unpatched code and passes with the fix:

• message-v2.test.ts: reasoning metadata survives cross-model sessions
• processor-effect.test.ts: trailing whitespace in reasoning text is preserved
• transform.test.ts: signed empty reasoning parts survive filtering
• transform.test.ts: unsigned empty parts still removed (existing behavior)

Full suite: 1867 pass, 0 fail.

Screenshots / recordings

Not a UI change.

Checklist

• I have tested my changes locally
• I have not included unrelated changes in this PR

[github-actions]

The following comment was made by an LLM, it may be inaccurate:

Based on my search, I found several related PRs addressing similar issues. Here are the most relevant ones:

Directly Related (mentioned in the PR description):

1. PR fix: preserve thinking block signatures and fix compaction headroom asymmetry #14393 - fix: preserve thinking block signatures and fix compaction headroom asymmetry #14393

   • "fix: preserve thinking block signatures and fix compaction headroom asymmetry"
   • Addresses cause 1 by removing the differentModel guard entirely. The current PR takes a narrower approach (reasoning only) and also covers causes 2 and 3.

2. PR fix(provider): skip empty-text filtering for assistant messages in normalizeMessages (#16748) #16750 - fix(provider): skip empty-text filtering for assistant messages in normalizeMessages (#16748) #16750

   • "fix(provider): skip empty-text filtering for assistant messages in normalizeMessages (normalizeMessages() removes empty text parts between reasoning blocks, invalidating Anthropic thinking block signatures #16748)"
   • Related to cause 3 (empty reasoning parts filtering)

Related to thinking block and reasoning content filtering:

3. PR fix(opencode): filter empty text content blocks for all providers #17742 - fix(opencode): filter empty text content blocks for all providers #17742

   • "fix(opencode): filter empty text content blocks for all providers"
   • Addresses similar empty content filtering issues

4. PR fix(provider): drop empty content messages after interleaved reasoning filter #17712 - fix(provider): drop empty content messages after interleaved reasoning filter #17712

   • "fix(provider): drop empty content messages after interleaved reasoning filter"
   • Related to empty reasoning parts handling

5. PR fix(provider): preserve assistant message content when reasoning blocks present #21370 - fix(provider): preserve assistant message content when reasoning blocks present #21370

   • "fix(provider): preserve assistant message content when reasoning blocks present"
   • Directly relevant to preserving reasoning block integrity

6. PR fix: strip reasoning parts when switching to non-interleaved models #11572 - fix: strip reasoning parts when switching to non-interleaved models #11572

   • "fix: strip reasoning parts when switching to non-interleaved models"
   • Related to model switching and reasoning content

The PR's description already acknowledges the relationship with #14393 and #12131, indicating these are complementary approaches to the same underlying issue.

[github-actions]

Thanks for updating your PR! It now meets our contributing guidelines. 👍