[BUG] --dangerously-skip-permissions: Plan mode only works on first invocation, subsequent uses skip planning and execute directly
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?
When running Claude Code with --dangerously-skip-permissions, plan mode (toggled via /plan) only activates correctly on the first use within a session. From the second invocation onward, plan mode is completely ignored — Claude Code skips the planning phase and directly executes actions.
What Should Happen?
Plan mode should activate consistently every time it is toggled on via /plan, regardless of how many times it has been used during the session. Claude Code should always present a plan for review before executing when plan mode is active.
Error Messages/Logs
No error messages displayed. The issue is behavioral — plan mode silently fails
to activate after the first use, with no warnings or errors in the output.
Steps to Reproduce
- Launch Claude Code with claude --dangerously-skip-permissions
- Toggle plan mode on via /plan
- Submit a prompt — Claude correctly enters plan mode and presents a plan ✅
- Toggle plan mode off, then toggle it back on via /plan
- Submit another prompt — Claude skips planning and executes directly ❌
Claude Model
Opus
Is this a regression?
Yes, this worked in a previous version
Last Working Version
2.1.75
Claude Code Version
2.1.91
Platform
Anthropic API
Operating System
Other Linux
Terminal/Shell
Xterm
Additional Information
This issue is 100% reproducible. The first /plan toggle in a session always works as expected, but every subsequent toggle fails silently. Restarting Claude Code resets the behavior (i.e., the first /plan works again, but the second does not).
This issue has 3 comments on GitHub. Read the full discussion on GitHub ↗