[BUG] Abnormal usage consumption - simple commands consuming 8-9% of 5-hour limit

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 today (March 23, 2026), token/usage consumption has become severely abnormal. Simple instructions that previously consumed 2-4% of the 5-hour limit are now consuming 8-9%. Even running claude --version consumed 3% of the limit.
Environment Info

Platform: Windows
Claude Code version: 1.2.4 (latest, confirmed with npm update -g @anthropic-ai/claude-code)
Model: claude-sonnet-4-6
Plan: Pro

Expected Behavior
A simple bug fix instruction should consume 2-4% of the 5-hour usage limit. Running claude --version should consume 0%.
Actual Behavior

2 simple instructions consumed 37% of the 5-hour limit
A subsequent bug fix instruction consumed 9% (38% → 47%)
Running claude --version alone consumed 3% (47% → 50%)
At this rate, only ~7 instructions are possible per 5-hour window instead of the usual 20+

Additional Context
I have been using Claude Code daily for over a month. This consumption pattern is completely new as of today. Sessions were fresh (new windows), not resumed. CLAUDE.md is 767 lines. No changes were made to my setup between yesterday and today.
This makes the service practically unusable at its current consumption rate.

What Should Happen?

A simple bug fix instruction should consume 2-4% of the 5-hour usage limit. Running claude --version should consume 0%.

Error Messages/Logs

Steps to Reproduce

Open a new Claude Code session
Send a simple bug fix instruction (no file uploads, no large context requests)
Observe usage meter

Claude Model

Sonnet (default)

Is this a regression?

Yes, this worked in a previous version

Last Working Version

_No response_

Claude Code Version

1.2.4

Platform

Other

Operating System

Windows

Terminal/Shell

VS Code integrated terminal

Additional Information

_No response_