[BUG] /usage: 5h block and weekly limit percentages rise in lockstep (should scale very differently)

Open 💬 1 comment Opened Jul 13, 2026 by DMC82-web

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?

During a single working session, the 5h block percentage and the weekly limit percentage in /usage rose together almost 1:1, when the weekly pool (aggregating ~33 5h blocks/week) should absorb the same token consumption much more slowly than the block.

What Should Happen?

The weekly percentage should increase much more slowly than the 5h block percentage for the same token consumption, since it is a much larger pool. It should never exceed the block's percentage from the same reference point.

Error Messages/Logs

None — this is a display/calculation anomaly in /usage, not an error.

Steps to Reproduce

  1. Note /usage block % and weekly % at the start of a working session.
  2. Work normally (this session included a deploy with a background subagent consuming ~54,500 tokens).
  3. Check /usage again at intervals through the session.

Readings observed:

  • Start: block 60% / weekly 60%
  • Mid-session: block 82% / weekly 83%
  • End of session: block 83% / weekly 84%

Claude Model

Sonnet (default)

Is this a regression?

I don't know

Last Working Version

_No response_

Claude Code Version

2.1.207

Platform

Other

Operating System

Windows

Terminal/Shell

PowerShell

Additional Information

Observed 2026-07-06, same session window as the /clear context-leak issue (cross-link the two after posting). Not certain if this is a display/calculation bug in /usage itself or an artifact of prior sessions/terminals that week having already consumed a large share of the weekly cap — flagging for investigation either way.

View original on GitHub ↗

This issue has 1 comment on GitHub. Read the full discussion on GitHub ↗