[BUG]

Environment

• Platform (select one):
• [x] Anthropic API
• Claude CLI version: <!-- output of claude --version -->
• Operating System: <!Ubuntu 22.04 -->
• Terminal: <!-- e.g. iTerm2, Terminal App -->

Bug Description

In simple terms, it does not adhere to the prompt. This is incredibly frustrating and costly.

Steps to Reproduce

1. <!-- First step -->
2. <!-- Second step -->
3. <!-- And so on... -->

Expected Behavior

<!-- What you expected to happen -->

Actual Behavior

In simple terms, it does not adhere to the prompt, even when I clearly include this - **DO NOT TAKE INITIATIVE**: Adhere to the instructions given in the prompt without engaging in any further actions. The model persistently and unceasingly takes initiative and produces poor code on top of more poor code. This is incredibly frustrating and costly.

Additional Context

<!-- Add any other context about the problem here, such as screenshots, logs, etc. -->