Digest / March 11, 2026
Codex Issues Digest — March 11, 2026
87 new issues filed ·
36 resolved in openai/codex (UTC day).
New issues (87)
- find_model_by_namespaced_suffix rejects hyphens in namespace, breaking custom provider slugs like llm-gateway/gpt-5.4
- --no-alt-screen still does not preserve scrollback in xterm.js-based terminals (VS Code integrated terminal reproduces)`
- Crash caused by missing pointer in underlying C++ code
- Support for Linux terminal(wsl) inside codex app (Windows).
- Allow queueing `/review` requests while agent performs work
- Regression in muli-agent spawning in 0.113.0
- The MCP Resources can often fill up the entire screen.
- Codex conversation titles sometimes contain stray symbols like 」}
- Error running remote compact task: timeout waiting for child process to exit
- Feature Request: Optional Prompt Optimization/Review Layer Before Execution Model
- 为什么新版 codexapp 在回复前总是执行 5 次 Reconnecting... 才开始回答，之前旧的版本就不会
- Codex hanging waiting for background script to finish executing (even though it seems to have)
- {"detail":"The 'gpt-5.3-codex' model is not supported when using Codex with a ChatGPT account."}
- Confusing unofficial Pypi codex-app-server-sdk
- Feature request: Support Markdown table rendering in Codex CLI
- Codex desktop app closes on startup when opening a specific Windows workspace, while CLI works normally
- Codex ran out of room in the model's context window. Start a new thread or clear earlier history before retrying.
- [codex-app-server Python SDK] UnicodeDecodeError with `Popen(text=True)` on non-ASCII Windows locales
- Agent stuck on "Waited for background terminal" after command finishes
- Support mouse event handling for seamless cursor movement in TUI
- [FEATURE] Allow disabling built-in skills via config.toml
- codex-cli and mac codex app often no react
- Race condition between subagent async notification and main agent turn lifecycle
- Codex CLI stuck at "thinking", `/status` shows "Limits: data not available yet" and account tier is "Unknown" after successful ChatGPT login (WSL)
- vscode extension make spring boot print filted debug log
- Repeated 'Error running remote compact task: timeout waiting for child process to exit'
- CODEX drained usage
- Chat layout does not resize correctly after exiting split-screen mode
- Codex repository size increased a lot due to accidental binaries commited recently
- Auth Problem
- Error: Missing optional dependency @openai/codex-darwin-x64 on macOS 12.7.6
- ignoring agent rules
- Team/Business accounts systematically excluded from every usage reset
- Environment Isolation in Codex for separate accounts and configs across projects and organizations
- The GPT-5.3-Codex model does NOT WORK in PAID ACCOUNT
- We Are the Lab Rats of OAI
- Better management of context output management structure
- Codex CLI believes in a `--skill` flag
- codex exec review ends without output after a fixed timeout and automation cannot distinguish the timeout
- Disable Codex prompting me to update npm package version
- "Skills" installed locations is missing .codex/skills
- Allow writable gitdir for current worktree in sandboxed workspace-write mode
- Clear context before implementing plan
- Windows CLI advertises `codex app`, but the `app` subcommand does not exist in codex-cli 0.114.0`
- GPT-5.4 frequently reports tasks as completed without executing them (returns previous outputs, false completion, or unperformed actions)
- Compacting is getting stuck
- Add --output-schema support to codex exec resume
- [possible degradation] Directories are now not trusted by default even with the `--dangerously-bypass-approvals-and-sandbox` option.
- Context Compaction Hanging
- Extend compaction prompt to reduce loss over multiple compactions
- SCAN protocol: preventing instruction drift in long sessions (~300 token overhead)
- Rate Limits dropped 30% within a few hours of use? Teams account - not getting usage resets and still getting fast drops
- Bad initial terminal path
- Intermediary status updates are buffered and appear only at end of turn
- Detect Git Worktree opened in another VSCode instance
- call_id exceeds 64-character limit when using non-OpenAI backends (Azure AI Foundry / LiteLLM)
- Add dedicated reasoning-depth hotkeys, non-semantic depth meter, and next-turn-safe switching
- Worker spawn inheritance broken: child worker agents do not inherit parent runtime/ network policy
- Add path-scoped review filters to codex exec review
- Add staged-only code review target via index snapshot
- Codex complicates a very simple directive
- Windows desktop app: packaged app-server exits with 0xC0000135 on launch, likely missing runtime dependency
- Remote compaction preserves assistant messages and leaves next turn at ~234k input tokens
- Windows: unified_exec PTY path bypasses Windows sandbox and can run unsandboxed despite workspace-write + network_access = false
- Codex macOS sandbox prevents Playwright browser tests from starting (bootstrap_check_in Permission denied (1100), listen EPERM 127.0.0.1)
- Add /release to show release notes
- Codex Desktop hides existing workspace threads on macOS
- Permissions error with git fsmonitor
- No answer from codex to my question
- code_mode is neither documented nor available in /experimental
- Plan mode does not ask for model spec
- Worker thread gets stuck on "Working" right before compacting context after tool call.
- Support click-to-copy and shift-click quote for structured reply sections
- Add a keyboard modal for selecting and actioning structured reply units
- Codex Desktop App does not display existing local sessions from shared CODEX_HOME
- Codex App: plan/checklist panel gets stuck on the first step and does not reflect completed tasks
- Shift input is silently dropped in Codex CLI in VS Code integrated terminal over WSL
- Provide the ability to allow workspace-write to access ./.codex
- Codex Desktop App — “failed to resume task” on all sessions after service incident
- Codex Fine-Grained Permissions
- Codex not starting on highly restricted network
- Infinite timeout on message q during compaction
- Updated and now I cannot pull up any threads, the thread keeps spamming like its refreshing and I have a red error box at the top
- Context corruption / attention issue
- windows sandbox: setup refresh failed with status exit code: 1
- Codex App on Windows intermittently redirects active threads to the "New Thread" page
- gpt-5.4 and gpt-5.3-codex unavailable via Codex extention on VSC and Codex App when using ChatGPT account
Resolved issues (36)
- Codex VS Code extension cannot authenticate when running on Remote SSH
- Allow queueing `/review` requests while agent performs work, when steering is on
- Support for Linux terminal(wsl) inside codex app (Windows).
- Codex conversation titles sometimes contain stray symbols like 」}
- VS Code extension crash can leave a thread permanently stuck in "thinking" with queued input
- VSCode Codex chat stuck on "thinking" after system crash
- stuck in "thinking" state in VS Code, unable to stop
- Codex Desktop threads stuck in “Thinking” and cannot be stopped unless background process is force-killed
- Prompt permanently stuck on "Thinking", even after pressing stop and restarting the Codex app
- 为什么新版 codexapp 在回复前总是执行 5 次 Reconnecting... 才开始回答，之前旧的版本就不会
- Confusing unofficial Pypi codex-app-server-sdk
- Feature request: Support Markdown table rendering in Codex CLI
- vscode extension make spring boot print filted debug log
- Chat layout does not resize correctly after exiting split-screen mode
- Codex CLI stuck at "thinking", `/status` shows "Limits: data not available yet" and account tier is "Unknown" after successful ChatGPT login (WSL)
- Auth Problem
- Clicking local file links with spaces opens a blank file in VS Code instead of the actual file
- ignoring agent rules
- We Are the Lab Rats of OAI
- Codex CLI believes in a `--skill` flag
- "Skills" installed locations is missing .codex/skills
- Disable Codex prompting me to update npm package version
- GPT-5.4 frequently reports tasks as completed without executing them (returns previous outputs, false completion, or unperformed actions)
- Intermediary status updates are buffered and appear only at end of turn
- {"detail":"The 'gpt-5.3-codex' model is not supported when using Codex with a ChatGPT account."}
- Rate Limits dropped 30% within a few hours of use? Teams account - not getting usage resets and still getting fast drops
- Unanswered questions following gpt-5.3-codex & gpt-5.4 restriction; is it permanent, why are paid users affected, and why was there no communication?
- CODEX drained usage
- Add /release to show release notes
- No answer from codex to my question
- code_mode is neither documented nor available in /experimental
- Worker thread gets stuck on "Working" right before compacting context after tool call.
- Experimental skills Github link is 404ing
- Codex repository size increased a lot due to accidental binaries commited recently
- Shift input is silently dropped in Codex CLI in VS Code integrated terminal over WSL
- Context corruption / attention issue