[BUG] Interrupting claude caused EVERY SINGLE PREVIOUS TASK COMPLETION NOTIFICATION TO EMIT and I couldn't stop it or even use `/feedback`

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?

I tried to interrupt claude but it caused EVERY SINGLE PREVIOUS TASK COMPLETION NOTIFICATION TO EMIT and I couldn't stop it - the session is not be recoverable because there's no checkpoint since the prior compaction:

```claude-code-session
⏺ Task(Fix cloud review feedback)
⎿  ❯ Read and execute: .ai/slash-command-phases/commit-push-review/prompts/implementation-subagent.md

⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review: API contract" completed
⎿  UserPromptSubmit hook error
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review: workslop detector" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review: mock demo data" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review: swiftdata migration" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review: accessibility compliance" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review: medical alert validator" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review: test coverage" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review: pr-toolkit code review" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review: feature-dev code review" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review: code quality" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review: documentation accuracy" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review: type design" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review: code simplicity" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review: performance" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review: silent failure hunter" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review: test analyzer" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review: comment analyzer" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review: naming quality" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review: security" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review iter2: pr-toolkit" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review iter2: simplicity" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review iter2: security" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review iter2: feature-dev" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review iter2: performance" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review iter2: test analyzer" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review iter2: naming" completed
⎿  UserPromptSubmit hook error
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review iter2: doc accuracy" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review iter2: comments" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review iter2: API contract" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review iter2: accessibility" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review iter2: medical alert" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review iter2: workslop" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review iter2: test coverage" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review iter2: type design" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review iter2: mock data" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review iter2: swiftdata" completed
⎿  Interrupted · What should Claude do instead?

⏺ Agent "Review iter2: code quality" completed
⎿  Context limit reached · /compact or /clear to continue

⏺ Agent "Review iter2: silent failures" completed
⎿  Context limit reached · /compact or /clear to continue

⏺ Agent "Review iter3: API contract" completed
⎿  Context limit reached · /compact or /clear to continue

⏺ Agent "Review iter3: feature-dev" completed
⎿  Context limit reached · /compact or /clear to continue

⏺ Agent "Review iter3: medical alert" completed
⎿  Context limit reached · /compact or /clear to continue

⏺ Agent "Review iter3: workslop" completed
⎿  Context limit reached · /compact or /clear to continue

⏺ Agent "Review iter3: simplicity" completed
⎿  Context limit reached · /compact or /clear to continue

⏺ Agent "Review iter3: pr-toolkit" completed
⎿  Context limit reached · /compact or /clear to continue

⏺ Agent "Review iter3: type design" completed
⎿  Context limit reached · /compact or /clear to continue

⏺ Agent "Review iter3: code quality" completed
⎿  Context limit reached · /compact or /clear to continue

⏺ Agent "Review iter3: security" completed
⎿  Context limit reached · /compact or /clear to continue

⏺ Agent "Review iter3: comments" completed
⎿  Context limit reached · /compact or /clear to continue

⏺ Agent "Review iter3: test analyzer" completed
⎿  Context limit reached · /compact or /clear to continue

⏺ Agent "Review iter3: naming" completed
⎿  Context limit reached · /compact or /clear to continue

⏺ Agent "Review iter3: performance" completed
⎿  Context limit reached · /compact or /clear to continue

⏺ Agent "Review iter3: accessibility" completed
⎿  Context limit reached · /compact or /clear to continue

⏺ Agent "Review iter3: doc accuracy" completed
⎿  Context limit reached · /compact or /clear to continue

⏺ Agent "Review iter3: test coverage" completed
⎿  Context limit reached · /compact or /clear to continue

⏺ Agent "Review iter3: mock data" completed
⎿  Context limit reached · /compact or /clear to continue

⏺ Agent "Review iter3: swiftdata" completed
⎿  Context limit reached · /compact or /clear to continue

⏺ Agent "Review iter3: silent failures" completed
⎿  Context limit reached · /compact or /clear to continue

❯ /feedback
╭─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╮
│ Submit Feedback / Bug Report │
│ │
│ Describe the issue below: │
│ │
│ Interrupting claude caused EVERY SINGLE PREVIOUS TASK COMPLETION NOTIFICATION TO EMIT and I couldn't stop it - session may not be recoverable because there's no checkpoint?! │
│ │
│ Could not submit feedback. Please try again later. │
│ │
│ Edit and press Enter to retry, or Esc to cancel │
│ │
╰─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯
Enter to continue · esc to cancel

I COULDN'T EVEN SUBMIT FEEDBACK! That's how broken it was!

### What Should Happen?

Uh, the already consumed outputs should just... not all stack immediately?

### Error Messages/Logs

```shell
`Could not submit feedback. Please try again later.`

Steps to Reproduce

1. stack up several dozen completion notifications by running a ton of subagents in a complex workflow
2. try to correct claude in some minor way by interrupting a task it's starting incorrectly
3. oooooopsies, you're out of luck! Every time a new notification arrives to the main context, it starts thinking about it. And every time you try to cancel it with ⎋, another shows up!

Claude Model

Opus

Is this a regression?

I don't know

Last Working Version

_No response_

Claude Code Version

2.1.14

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

Terminal.app (macOS)

Additional Information

Okay, I mean, 2.1.14 isn't the latest, but I don't think the react compiler will fix this.