Digest / November 18, 2025
Codex Issues Digest — November 18, 2025
14 new issues filed ·
20 resolved in openai/codex (UTC day).
New issues (14)
- Expand MacOS Seatbelt-exec sandbox with network_access = true to restrict to localhost
- 'Open config.toml in WSL environment' broken (incorrect path)
- Connection failed: error sending request for url (https://ai.megallm.io/v1/ chat/completions?)
- PRO user using codex being told to f off by gpt-5.1 over and over and will not do anything in agent mode.
- Real-Time Steering During Codex Runs (Similar to Claude Code)
- Auto complete / Code completion
- Document the Landlock requirement for Linux
- Easier deletion of file paths in prompt
- [WireApi::Chat] Token usage not updated correctly, causing infinite function_call loop
- Significant Performance Regression in Codex from v0.4.39(VSCode) to v0.4.40+(VSCode) on Specific Repository
- TUI feedback always links to bug-report template even for good result/other
- Ahead of time (background) context compaction to hide latency
- OAuth Login Fails Behind Corporate Proxy with Custom CA Certificates
- Garbled output / not setting dumb terminal mode with shell snapshot on
Resolved issues (20)
- Stream disconnected before completion
- Input exceeds context window error
- “stream disconnected before completion” + “input exceeds context window” loop
- Codex extension in VSCode: Completely ignores "Allow every time", no matter how many times I click it - And despite setting to Agent (full access)
- Expand MacOS Seatbelt-exec sandbox with network_access = true to restrict to localhost
- 🖐 stream disconnected before completion: stream closed before response.completed
- all commands in sandbox mode fail and need approval
- Codex extension in vscode: MCP server (Context7) fails with “request timed out”
- Codex cannot login to chatgpt plus account if created with email/password and then used google auth
- Connection failed: error sending request for url (https://ai.megallm.io/v1/ chat/completions?)
- PRO user using codex being told to f off by gpt-5.1 over and over and will not do anything in agent mode.
- Stream disconnected before completion: Transport error
- Token exchange failed: token endpoint returned status 403 Forbidden
- Codex ran out of room in the model's context window. Start a new conversation or clear earlier history before retrying.
- rmcp OAuth tokens never refresh based on expires_at
- stream error: unexpected status 404 Not Found: {"detail":"Not Found"}
- 🐛 Mouse / scroll in zellij doesn’t let me view earlier TUI conversation (alt-screen interaction)
- You've hit your usage limit. Try again in 4 days 20 hours 9 minutes.
- Quota exceeded. Check your plan and billing details.
- Can't log in to a paid Team workspace