[BUG] “I am being charged for actions I cannot observe, control, or attribute to my intent.”

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?

Claude Code appears to consume a significant portion of session usage for actions that are not visible to the user (e.g., implicit file reads, context loading, or tool calls).
This makes it difficult to understand or predict usage, even for simple prompts, and breaks the expectation that cost should scale with user intent.
Usage increases significantly even when:

• prompts are minimal
• no files are explicitly referenced
• no exploration is requested

This suggests that background or implicit actions are consuming quota without any visibility or attribution to specific user actions.

What Should Happen?

Usage should reflect only user-intended actions, scale with prompt complexity, and provide sufficient visibility (e.g., per-prompt or per-action breakdown) so users can understand and verify how their quota is consumed.

Error Messages/Logs

Steps to Reproduce

Open Claude Code and continue using an existing session (e.g., from a previous day).
Ensure no files are explicitly opened or referenced in the prompt.
Submit 1–2 simple prompts (e.g., short questions or trivial code like print("hello")).
Observe the session usage indicator after each prompt.

Claude Model

None

Is this a regression?

Yes, this worked in a previous version

Last Working Version

_No response_

Claude Code Version

2.1.100 (Claude Code)

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

Terminal.app (macOS)

Additional Information

_No response_