Digest / February 13, 2026
Codex Issues Digest — February 13, 2026
62 new issues filed ·
32 resolved in openai/codex (UTC day).
New issues (62)
- Cannot unbind cmd+shift+P "enable plan" command?
- Typing on the CLI since 0.100 is very slow
- Stream disconnect when working with spark model
- Codex spark timeout
- Regression: stream disconnected before completion on Windows in 0.100.0+ (works in 0.99.0)
- AUTO ACCEPT COMMANDS IN CLI PLSSS
- Codex CLI Issue Not Determining Account Type
- Unable to resume large conversations
- Unexpected Bengali-like character injection in ko/en output since Codex 5.3
- TUI banner shows v0.98.0 after upgrading to 0.101.0
- JetBrains (IntelliJ IDEA) ACP / Codex: AI Chat becomes unusable after using gpt-5.3-codex; session/resume fails with “no rollout found for thread id …” and persists even after switching back to gpt-5.2-codex
- Consider Introducing Read–Write Locks to Enhance Concurrent Work
- Allow overriding the base URL for OpenAI in config.toml
- Subagent configuration and orchestration
- Create branch from chat so we can work multiple things in parallel with shared context.
- add “rename session” please
- Shortcut description to paste image
- /skills updates are not reflected in model-visible “Available skills” within the same CLI session
- Infinite Loading after Codex extension installation
- Unable to view the latest Codex 5.3 model
- Add Config Option to Disable Automatic Context Compaction in Codex CLI
- App seems to use the envrionment file from the main project no matter what branch is selected
- History (chats) gone after logging in again (API)
- Group skills by scope in the Skills page
- Cannot create new local conversation (WSL)
- Auto-compact fails
- “Limit reached” banner persists while using credits; remaining credit not displayed
- "Failed to resume task" when trying create "New Codex Agent"
- App-server config write APIs cannot target project .codex/config.toml (user config only)
- Question: More details and context to facilitate decisions
- Use dash-separated header session-id instead of session_id
- Does not register use all the time command
- gpt-5-3-codex model not available anymore in list
- Stream disconnected before completion
- Feature request: auto-plan mode with automatic recommended choices
- VS Code Codex extension hangs after completing edits and keeps CMD running with high CPU
- [PDF Skill] macOS: dependency install guidance should default to venv, not system `python -m pip`
- stream disconnected before completion: error sending request for url (https://chatgpt.com/backend-api/codex/responses)
- Add rewind feature
- Stream disconnected before completion on VPS using ChatGPT OAuth (worked 12 hours prior, reproducible, network verified)
- npm install -g @openai/codex@latest broken on Windows after dist-tag packaging change (#11339)
- Feature Request: Add `--add-dir` flag to include additional working directories
- Add `/add-dir` slash command to add directories mid-session
- cannot open any new task for specific project, giving "Error loading task"
- Add `codex exec fork` subcommand for headless/non-interactive fork
- Todo list above the input line
- Have @codex engage in all GitHub issue threads
- Compressing context BUG
- Desktop rehydrate injected stale AGENTS instructions from another workspace
- Plan downloads in the VS code extension should save the plan to the repo
- VSCode Tab Completions w/Spark
- Approval popup single-key shortcuts (y/n) capture keystrokes while typing follow-up prompt
- experimental feature "collab" is broken on main for API/free/plus/go users
- Manage MCP servers UX
- Feature request – automatic orchestration from a plan file (Boss+ Workers)
- Push to talk binding for App
- 2x Rate Limits not implemented
- Codex is not using credits after weekly limit is reached (Team Pro plan)
- Usage Command No Longer Available
- Codex cannot handshake with WebStorm MCP server (SSE endpoint), initialize fails
- Codex context persistence issue after closing folders in VS Code
- Allow configuring subagent model and reasoning_effort in config
Resolved issues (32)
- Codex 5.3 model not doing any file edits within CLI
- codex-rs: grep_files tests fail due unknown test-model fallback; unified_exec test races end event
- AUTO ACCEPT COMMANDS IN CLI PLSSS
- Unexpected Bengali-like character injection in ko/en output since Codex 5.3
- Can't find Codex Spark in the model selection
- Codex CLI prompts API users to break their config (to use an unsupported model)
- Create branch from chat so we can work multiple things in parallel with shared context.
- Cannot upload images anymore
- Error: "state db missing rollout path" received when running codex exec resume <SESSION_ID> "..."
- Codex app on MacBook-pro (M4-max) - Massive CPU usage on startup for 55 seconds + Massive performance issues and lag + Response breaking
- Feature Request: Add `--add-dir` flag to include additional working directories
- Codex exited after repeated stream disconnects; now I can’t use gpt-5.3-codex
- add “rename session” please
- Codex Desktop App: gpt-5.3-codex not shown in model selector dropdown
- Codex over SSH drops keystrokes, emits IO errors.
- Add Config Option to Disable Automatic Context Compaction in Codex CLI
- Auto-compact fails
- Codex is not able to write to repo root .agents directory
- Codex CLI Issue Not Determining Account Type
- Stream disconnected before completion on VPS using ChatGPT OAuth (worked 12 hours prior, reproducible, network verified)
- stream disconnected before completion: error sending request for url (https://chatgpt.com/backend-api/codex/responses)
- Unable to view the latest Codex 5.3 model
- Compressing context BUG
- VSCode Tab Completions w/Spark
- Approval popup single-key shortcuts (y/n) capture keystrokes while typing follow-up prompt
- Add rewind feature
- experimental feature "collab" is broken on main for API/free/plus/go users
- MCP Initialization Issue
- Rule syntax error on individual file invalidates all the `.rules` files.
- Usage Command No Longer Available
- Codex context persistence issue after closing folders in VS Code
- Allow configuring subagent model and reasoning_effort in config