fix(opencode): repair thinking blocks rejected by Anthropic API

[iceteaSA]

Problem

Anthropic signs each thinking block against its original position. Two things break signatures and trigger thinking blocks cannot be modified 400s:

1. Interleaved thinking flattened by the AI SDK — opencode's Vercel AI SDK regroups all thinking blocks to the front of the assistant turn instead of leaving them interleaved with the tool_use blocks they authorized. The displaced blocks no longer match their signed positions.

2. Stale signatures on historical assistant turns replayed in a new request context (after context compaction or provider failover).

3. Final block constraint — Anthropic also rejects an assistant message whose final block is a thinking block.

Strategy

• Keep a single signed thinking block only when it is the lone block in the latest assistant turn (the normal, still-valid current-turn case).
• Downgrade all others to <thinking> text — preserves reasoning content without invalid signatures.
• Drop redacted_thinking blocks (no readable text to preserve).
• Guard against a trailing thinking block being the final block in the message.
• Sanitize lone UTF-16 surrogates in downgraded text to avoid invalid-UTF8 400s.

Cache impact

None on the static prefix. The transform runs before cache_control anchors are applied, so anchors land on the already-normalized structure. Historical turns render deterministically once they become non-last. The only cost is the unavoidable one-time tail transition when the current turn becomes historical.

Tests

6 new test cases covering: flattened interleaving, single-block preservation, historical downgrades, redacted_thinking drops, lone-surrogate sanitization, and trailing-thinking-block downgrade.

---

^{Need help on this PR? Tag @codesmith with what you need. Autofix is disabled.}

---

Summary by cubic

Repairs Anthropic 400s by normalizing thinking blocks in opencode requests. Prevents errors from flattened interleaving, stale signatures on historical turns, and final thinking blocks.

• Bug Fixes
  • Keep a single signed thinking block only in the latest assistant turn; downgrade all others to <thinking> text.
  • Drop redacted_thinking blocks unless it’s the single block in the latest assistant turn; never allow it as the final block.
  • Prevent thinking as the final block in an assistant message.
  • Sanitize lone UTF-16 surrogates in downgraded text.

^{Written for commit b6971c4. Summary will update on new commits.}

[cubic-dev-ai]

No issues found across 2 files

_{Re-trigger cubic}

[ualtinok]

Closing this in favor of the upstream OpenCode fix: anomalyco/opencode#30182.

This PR repairs the final Anthropic payload by downgrading/dropping thinking / redacted_thinking blocks. That can avoid 400s, but it is a broad and destructive sanitizer: it can discard valid signed/redacted thinking that should be preserved when the upstream message shape is correct.

The upstream PR addresses the signed-thinking reorder issue at the source by keeping OpenCode's existing ordinary-text reorder while skipping it when it would move signed or redacted Anthropic reasoning.

After that lands, we should update the OpenCode dependency and re-test. If there is still a concrete residual invalid-body case, we can open a narrower plugin-side guard for that specific shape.