I was told to compact to save usage, and it consumed all my usage before I could do anything

Resolved 💬 2 comments Opened Apr 24, 2026 by muncrief Closed May 28, 2026

Preflight Checklist

  • [x] I have searched existing issues for similar behavior reports
  • [x] This report does NOT contain sensitive information (API keys, passwords, etc.)

Type of Behavior Issue

Other unexpected behavior

What You Asked Claude to Do

I entered Claude this afternoon with "claude --resume" and was immediately presented with a message that said I should compact to save usage. I was presented with a numbered list with 1 being agreeing to the compaction. So I entered 1 and Claude started compaction. Then after about 15 minutes I was suddenly presented with a message that said my usage limit had been reached and I would have to wait until 5:50 PM before I could continue. And then I was asked if I wanted to upgrade my Pro plan or wait. So this was obviously a scam to try and get me to give Anthropic more money.

This was a really rotten thing to do. Compaction was prompted as a way to save usage while in truth it consumed my entire remaining quota with no cost warning. I want an explanation from Anthropic explaining why they did this, and a refund of my tokens which were taken under false pretense.

What Claude Actually Did

It stole my usage tokens before I'd done anything other than agree to what it recommended.

Expected Behavior

I expected it would do what I said - compact to save usage. Not steal my tokens under false pretense.

Files Affected

Permission Mode

Accept Edits was ON (auto-accepting changes)

Can You Reproduce This?

No, only happened once

Steps to Reproduce

_No response_

Claude Model

Sonnet

Relevant Conversation

Claude recommended I compact to save usage and when I agreed it stole all my tokens.

Impact

Critical - Data loss or corrupted project

Claude Code Version

2.1.118 (Claude Code)

Platform

Anthropic API

Additional Context

_No response_

View original on GitHub ↗

This issue has 2 comments on GitHub. Read the full discussion on GitHub ↗