Digest / January 28, 2026
Codex Issues Digest — January 28, 2026
47 new issues filed ·
45 resolved in openai/codex (UTC day).
New issues (47)
- Multiple Nix build failures
- Terminal title bar
- Is there a way to use the GPT-5.2 Pro model directly in Codex CLI?
- The `exec --yolo` mode cannot run `rm -rf`.
- CLI cancelling agent when no keyboard input
- double final message
- Getting rate limited for a Pro user despite plenty of usage left
- Rate limit reached with chatgpt sub
- Hitting TPM limit
- Stream disconnected before completion due to TPM rate limit (250k TPM exceeded)
- Stream disconnected before completion: Rate limit reached for organization
- Codex CLI blocked by rate limits for unrelated org
- Rate limit reached for organization on tokens per min (TPM)
- `stream disconnected before completion` on `RPM`
- Rate limit reached for organization org- on requests per min (RPM): Limit 3000, Used 3000, Requested 1
- codex responses with 'Rate limit reached for organization'
- TPM ISSUE
- [Bug] Pasting multiline text triggers auto-submit and strips newlines (v0.89.0 - 0.92.0)
- codex revert and timeline
- Feature: Add --agents flag to switch between named AGENTS.md variants
- use brew install codex on apple arm64 install x86_64 codex binary, should install arm64 codex.
- Visual bug when stopping a MCP in mid call
- timeout waiting for child process to exit
- Tool 'web_search_preview' is not supported with gpt-5.1-codex-max-2025-12-04.
- Running process names are cut off
- Support /personality for remote models
- codex can not auto run go test with chromedp
- Codex Teste 1
- Codex hangs
- `GPT-5.2-High` feels off
- Notify user when Codex needs attention
- please add fallback API key feature
- Feature request: Custom conversation titles + checkpoint rollback / branching in Codex Extension chat
- Exception setting "CursorPosition": "The handle is invalid."
- Plan mode offer 4 bad options and no way to say to model my custom one if all are bad.
- `elevated_windows_sandbox` causing all agent commands to fail with `(no output)` (logs show `CreateProcessAsUserW failed: 5`)
- Search chat contents
- Codex v0.92.0; VSCode extension; missing 'settings' field
- Rate Limits crash codex
- VS Code OpenAI codex keeps asking for permissions: Approve Once, Approve This Session, Reject.
- Introduce equivalent of Copilot's `#githubRepo` context
- Support `notifications/tools/list_changed`
- non-interactive support for fork
- MDM docs mention non-existant `codex config` command
- Codex CLI v0.92.0 crashes on startup with ZodError (tools[5].type invalid)
- Codex 0.92.0: web_search tool defined even when web search is disabled.
- MCP resume_path for Codex sessions
Resolved issues (45)
- WSL: shortcut overlay snapshot tests are non-deterministic
- Group or Filter Chat History by Project Context
- Terminal title bar
- The `exec --yolo` mode cannot run `rm -rf`.
- Hitting TPM limit
- Stream disconnected before completion due to TPM rate limit (250k TPM exceeded)
- Rate limit reached for organization on tokens per min (TPM)
- Rate limit reached for organization org- on requests per min (RPM): Limit 3000, Used 3000, Requested 1
- Getting rate limited for a Pro user despite plenty of usage left
- Stream disconnected before completion: Rate limit reached for organization
- Codex CLI blocked by rate limits for unrelated org
- `stream disconnected before completion` on `RPM`
- codex responses with 'Rate limit reached for organization'
- Rate limit reached with chatgpt sub
- unable to use GPT-5.2
- codex revert and timeline
- TPM ISSUE
- Incorrect tooltip: functionality advertised is disabled by default
- #9954 don't work
- new release (codex on cloud) has significant regression with context retention and code generation
- Performance: Browser becomes unresponsive with very large answers (>5,000 lines)
- Codex hangs repeatedly when "Updating Python path for module import"
- Codex CLI hangs indefinitely when the workspace is out of credits
- [linux sandbox] failed socket write hangs python asyncio
- calling codex from the current openai-agents produces spurious RPC errors `Elicitation not supported`
- gpt-5-codex-mini does not work with api-key
- Never finish the task
- Codex Extension will not / almost never completes a single prompt on new update
- There is a problem inputting Chinese in a terminal created by screen
- Low-credit picker interrupts Codex CLI workflows
- Regression: v0.4.35 Correctly Identifies Debian WSL but Later Version Require Ubuntu
- [Windows] clipboard paste drops characters, Shift+Enter and Ctrl+Enter insert [13;2u instead of a newline
- Support /personality for remote models
- Codex Teste 1
- please add fallback API key feature
- Notify user when Codex needs attention
- Feature request: Custom conversation titles + checkpoint rollback / branching in Codex Extension chat
- Search chat contents
- [Bug] Pasting multiline text triggers auto-submit and strips newlines (v0.89.0 - 0.92.0)
- Exception setting "CursorPosition": "The handle is invalid."
- Plan mode offer 4 bad options and no way to say to model my custom one if all are bad.
- Codex hangs
- Rate Limits crash codex
- MDM docs mention non-existant `codex config` command
- use brew install codex on apple arm64 install x86_64 codex binary, should install arm64 codex.