[BUG] stream-json output mode no longer includes thinking blocks since v2.1.8

Preflight Checklist

• [x] I have searched existing issues and this hasn't been reported yet
• [x] This is a single bug report (please file separate reports for different bugs)
• [x] I am using the latest version of Claude Code

What's Wrong?

Since version 2.1.8, the --output-format stream-json mode no longer outputs type: "thinking" content blocks.

What Should Happen?

The stream-json output should include thinking blocks like it did in v2.1.7 and earlier:

{"type":"assistant","message":{"content":[{"type":"thinking","thinking":"The user is asking...","signature":"..."}]}}

Error Messages/Logs

Steps to Reproduce

1. Install Claude Code v2.1.8 or later
2. Run with stream-json output:

claude -p "Ultrathink: what is 2+2?" --output-format stream-json --verbose

1. Observe output contains only {"type":"text","text":"4"}` - message, no thinking block
2. Compare with v2.1.75:

npx @anthropic-ai/claude-code@2.1.7 -p "Ultrathink: what is 2+2?" --output-format stream-json --verbose

1. Observe output includes thinking message {"type":"thinking","thinking":"...","signature":"..."}

Claude Model

Opus

Is this a regression?

Yes, this worked in a previous version

Last Working Version

2.1.7

Claude Code Version

2.1.15 (Claude Code)

Platform

Anthropic API

Operating System

Other Linux

Terminal/Shell

Terminal.app (macOS)

Additional Information

_No response_