ability to auto accept when implementing a big feature instead of keeping hitting accept.

Resolved 💬 5 comments Opened Nov 16, 2025 by GhouI Closed Nov 23, 2025

💡 Likely answer: A maintainer (github-actions[bot], contributor) responded on this thread — see the highlighted reply below.

What feature would you like to see?

I am currently implementing a feature. I want the bot to stop asking for confirmation for commands, but there should still be an option in the middle of a conversation to activate a dangerous mode without repeatedly selecting 'Yes, proceed'

Additional information

_No response_

View original on GitHub ↗

5 Comments

github-actions[bot] contributor · 9 months ago

Potential duplicates detected. Please review them and close your issue if it is a duplicate.

#5112
#5157
#5199
#6112
#6542

Powered by Codex Action

etraut-openai contributor · 9 months ago

Both the CLI and the IDE extension allow you to specify "Full Access" mode, which allows you to run without sandboxing. In the CLI, use the /approvals slash command to set this mode. In the IDE Extension, use the picker beneath the prompt input box. We recommend using this mode only inside of an isolated environment (a docker container or VM, for example).

GhouI · 9 months ago

Both the CLI and the IDE extension allow you to specify "Full Access" mode, which allows you to run without sandboxing. In the CLI, use the /approvals slash command to set this mode. In the IDE Extension, use the picker beneath the prompt input box. We recommend using this mode only inside of an isolated environment (a docker container or VM, for example).

/approval can't run when the agent is working. is it possible to allow this mid run?

etraut-openai contributor · 9 months ago

@GhouI, that feature is captured by https://github.com/openai/codex/issues/4394. Please upvote if you'd like us to prioritize it.

GhouI · 9 months ago

Thank you I have upvoted it.