[BUG] Model incorrectly believes it's in interactive mode when using -p flag

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?

When running Claude Code in non-interactive print mode (-p), the model believes it's running interactively. This prevents users from providing mode-specific instructions (e.g., in CLAUDE.md) that should only apply to -p mode output.

What Should Happen?

The model should be aware it's running in -p mode, either via system prompt context or environment information.

Error Messages/Logs

Steps to Reproduce

$ claude -p 'Are you running interactively or non-interactively (in -p mode)?'
I'm running interactively. You're communicating with me through an interactive CLI session, not in non-interactive `-p` (print) mode. In `-p` mode, I would receive a prompt and produce output without back-and-forth conversation, but here we can have a dialogue.

Claude Model

Opus

Is this a regression?

I don't know

Last Working Version

_No response_

Claude Code Version

2.1.5 (Claude Code)

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

Other

Additional Information

I ran into this because I wanted to provide conditional instructions like "when in -p mode, output strict CommonMark markdown", to work around #17554 (which I only care about when I pipe the output to pandoc)