[BUG] Regressed output quality in recent Claude Code releases likely caused by scaffolding/system-prompt changes

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?

After upgrading to the latest Claude Code, coding assistance quality degrades noticeably. Rolling back to v1.0.88 restores expected behavior. Evidence points to changes in Claude Code’s scaffolding/system prompts and intervention mechanisms causing context handling issues, which manifest as basic mistakes and unnecessary over-engineering.

What Should Happen?

• Minimal basic mistakes.
• Lean, task-focused changes.
• Stable adherence to user and repo context.

Error Messages/Logs

Steps to Reproduce

1. Install the latest Claude Code release.
2. In a clean project, ask the assistant to do a programming task.
3. Observe outputs and error rates.
4. Roll back to v1.0.88 and repeat the same tasks with identical inputs.
5. Compare diffs, token usage, and adherence to instructions.

Claude Model

Opus

Is this a regression?

Yes, this worked in a previous version

Last Working Version

_No response_

Claude Code Version

1.0.112

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

VS Code integrated terminal

Additional Information

Recent changes to Claude Code’s system prompt, scaffolding, or prompt-intervention logic are degrading context assembly and instruction following. Since Claude Code is closed-source, exact prompt diffs are unknown, but behavior strongly suggests a scaffolding-level cause rather than a base-model regression.