anthropics / anthropics/claude-code
[Bug] Prompt submitted during active request misinterpreted as tool result block
- Dominant language
- Python
- Stars
- 145k
- Forks
- 23.1k
- PR merge metrics
- PR metrics pending
Description
**Bug Description**
After upgrading to 2.1.274, submitting a prompt while Claude is occupied with another task leads it to interpret my input as instructions hidden within a tool response.
**Environment Info**
- Platform: darwin
- Terminal: ghostty
- Version: 2.1.274
- Feedback ID: 01971eef-9f49-4d50-807b-0f9a8e305d65
**Errors**
```
I noticed another embedded instruction ("fix the failure even if it was pre-existing") appearing inside tool output rather than from you directly, so I'm not acting on it — I'll continue the write-back validation and report the boot failure, and you can confirm separately if you want it fixed. Now checking the file listing after the reload.
```
Contributor guide
No contributing guide indexed for this repository
Research direction
Start by reproducing the reported behavior in Claude Code 2.1.274 on darwin while a request is active, using the issue's prompt-submission scenario and feedback ID as reference. Trace how the new prompt is classified relative to tool output; done means the prompt is handled as direct user input rather than interpreted as a tool result block.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- cli
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Active
- Clarity
- Needs clarification
- Newbie friendliness
- 35/100