[Bug] Agent incorrectly interprets injected "user " prefixed messages as user input

Status Open
Reported on v2.1.220
Maintainer reply None cached
Activity 1 comment · opened Jul 31, 2026

Bug Description
Claude code keeps injecting messages starting with "user ". It confiuses the agent thinking I'm saying this I'm not saying.

Environment Info

  • Platform: linux
  • Terminal: ghostty
  • Version: 2.1.220
  • Feedback ID: b35b219e-8e49-4a5b-8a63-97bbef3d6b4c

Errors

[]

View original on GitHub ↗

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