[BUG] Claude Code input unresponsive — "Usage limit reached" despite being well under 5-hour (28%) and weekly (16%) limits on Max 20x plan

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 has become completely unusable. The input prompt is unresponsive — I cannot type anything into it at all. The status bar reads "Usage limit reached" and the input area label is "Ask permissions".

However, my in-app plan usage panel shows I'm nowhere near any limit:

• 5-hour limit: 28% used (resets in ~2h)
• Weekly · all models: 16% used (resets in 5d)
• Weekly · Claude Design: 0%
• Sonnet only: 0%

I am on the Max 20x plan (highest individual subscription tier), so the displayed usage should allow me to continue. The enforcement and the displayed usage clearly disagree.

I also contacted the in-app Fin AI support agent, which offered to escalate to the team, then said it couldn't actually escalate. This left me with no working path to report the issue except via this GitHub repo, since the /bug command inside Claude Code is also unusable (the input is frozen).

What Should Happen?

Since I am well under both the 5-hour (28%) and weekly (16%) limits on a Max 20x plan, Claude Code should continue to accept prompts normally. If I had genuinely hit a limit, the displayed usage should reflect that (e.g. ~100%), and the message should be accurate and actionable. The usage enforcement and the displayed usage should be consistent.

Error Messages/Logs

Status bar shows: "Usage limit reached"
Input field label: "Ask permissions"
Session header line (example): "Ready  Test communication connection   akoti  10m"

Steps to Reproduce

1. Use Claude Code normally for routine coding work (enhancing an existing application, no unusually heavy requests).
2. At around 18:05 local time (BST), the input prompt stops accepting any keystrokes.
3. Status bar switches to "Usage limit reached".
4. Open plan usage panel — shows 28% 5-hour, 16% weekly, plenty of headroom.
5. No ability to type anything, including /bug, so the in-app reporting path is blocked.

Claude Model

None

Is this a regression?

Yes, this worked in a previous version

Last Working Version

2.1.112 (same version that is now broken; worked earlier today until ~18:05 BST)

Claude Code Version

2.1.112 (Claude Code)

Platform

Anthropic API

Operating System

Windows

Terminal/Shell

Windows Terminal

Additional Information

Environment details:

• OS: Windows 11 Home, build 26200
• Terminal: Windows Terminal running PowerShell (WSL installed but not used for this session)
• Claude Code install: dual install via npm and winget — npm wins the PATH race (the active binary comes from the npm install)
• Plan: Max 20x
• Timezone: BST (Europe/London)

Extra context:

• Interaction with the in-app Fin AI support agent was unhelpful. Fin first offered to escalate to the team, then said it cannot escalate. This is a separate UX issue worth noting, because for users who cannot use Claude Code at all, this chat path is the only one offered in-app, and it terminates in a dead end.
• Suggested investigation: potential sync/mismatch between the usage-calculation service that enforces "Usage limit reached" and the usage-display service shown in the plan panel. It's also possible that a request may have spiked the enforcement counter beyond what the display shows.
• I'd appreciate either a fix, a concrete workaround, or visibility into which limit was actually tripped so I can correlate with my activity.