Skip to content

fix(codex): adapt agent cycle for chatgpt auth and unattended execution#13

Open
paknahad wants to merge 1 commit intonkhdiscovery:mainfrom
paknahad:fix/codex-agent
Open

fix(codex): adapt agent cycle for chatgpt auth and unattended execution#13
paknahad wants to merge 1 commit intonkhdiscovery:mainfrom
paknahad:fix/codex-agent

Conversation

@paknahad
Copy link
Copy Markdown
Contributor

@paknahad paknahad commented May 2, 2026

  • Added is_chatgpt_auth to detect ChatGPT web session authentication.
  • For ChatGPT auth, run codex without a model argument (as it defaults to the account model) in an isolated temporary CODEX_HOME.
  • Replaced --full-auto with --dangerously-bypass-approvals-and-sandbox for non-interactive execution, since the agent runs in a Docker container.
  • Redirect stdin from /dev/null to prevent the Codex CLI from hanging.

- Added is_chatgpt_auth to detect ChatGPT web session authentication.
- For ChatGPT auth, run codex without a model argument (as it defaults to the account model) in an isolated temporary CODEX_HOME.
- Replaced --full-auto with --dangerously-bypass-approvals-and-sandbox for non-interactive execution, since the agent runs in a Docker container.
- Redirect stdin from /dev/null to prevent the Codex CLI from hanging.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant