›_ Codex Issues
Codex Issues / #11568

VSCode extension crashes

Open 💬 8 comments Opened Feb 12, 2026 by tkysB
💡 Likely answer: A maintainer (etraut-openai, contributor) responded on this thread — see the highlighted reply below.

What version of the IDE extension are you using?

Codex extension: 0.4.51+ (regression starts at 0.4.51)

What subscription do you have?

ChatGPT plus -> pro (upgraded 1 day ago)

Which IDE are you using?

vs code

What platform is your computer?

Windows 11 pro

What issue are you seeing?

Since the Codex version exceeded 0.4.51, my Codex has completely stopped working. Even when I run very simple requests that require basic tool usage such as “Please check AGENTS.md” or “Please check the current time”, the error below occurs and Codex crashes.

I consulted ChatGPT and tried various configuration changes, including reinitializing ~/.codex, but nothing has resolved the issue. I want to use 5.3-codex.

--- Error Message
エラーが発生しました
Codex が以下のエラーによって停止しました：

Codex process errored: Codex process is not available
2026-02-12T09:26:15.816302Z ERROR codex_core::rollout::recorder: Falling back on rollout system
2026-02-12T09:26:15.816333Z  WARN codex_core::state_db: state db record_discrepancy: list_threads_with_db_fallback, falling_back
---

--- vs code Log
2026-02-12 18:19:09.918 [info] Skills/list request cwdsCount={} forceReload={} templateArgs=[1,false]
2026-02-12 18:19:10.372 [info] Skills/list ok entries={} templateArgs=[1]
2026-02-12 18:24:33.531 [info] Skills/list request cwdsCount={} forceReload={} templateArgs=[1,false]
2026-02-12 18:24:33.763 [info] Skills/list ok entries={} templateArgs=[1]
2026-02-12 18:26:09.442 [error] [CodexMcpConnection] cli:} message="2026-02-12T09:26:09.442033Z  WARN codex_core::shell_snapshot: Failed to create shell snapshot for powershell: Shell snapshot not supported yet for PowerShell\n\nStack backtrace:\n   0: <unknown>\n   1: <unknown>\n   2: tree_sitter_bash\n   3: tree_sitter_bash\n   4: <unknown>\n   5: <unknown>\n   6: <unknown>\n   7: <unknown>\n   8: <unknown>\n   9: <unknown>\n  10: <unknown>\n  11: <unknown>\n  12: <unknown>\n  13: BaseThreadInitThunk\n  14: RtlUserThreadStart"
2026-02-12 18:26:09.702 [warning] [CodexMcpConnection] cli:} message="codex_core::turn_metadata: timed out after 250ms while building turn metadata header"
2026-02-12 18:26:12.677 [info] Conversation created: {} templateArgs=["019c512c-151c-79b2-9f45-cbcd3bbc5409"]
2026-02-12 18:26:12.956 [warning] [CodexMcpConnection] cli:} message="codex_protocol::openai_models: Model personality requested but model_messages is missing, falling back to base instructions. model=gpt-5.1-codex-mini personality=pragmatic"
2026-02-12 18:26:12.962 [error] [CodexMcpConnection] cli:} message="2026-02-12T09:26:12.961528Z  WARN codex_core::shell_snapshot: Failed to create shell snapshot for powershell: Shell snapshot not supported yet for PowerShell\n\nStack backtrace:\n   0: <unknown>\n   1: <unknown>\n   2: tree_sitter_bash\n   3: tree_sitter_bash\n   4: <unknown>\n   5: <unknown>\n   6: <unknown>\n   7: <unknown>\n   8: <unknown>\n   9: <unknown>\n  10: <unknown>\n  11: <unknown>\n  12: <unknown>\n  13: BaseThreadInitThunk\n  14: RtlUserThreadStart"
2026-02-12 18:26:13.219 [warning] [CodexMcpConnection] cli:} message="codex_core::turn_metadata: timed out after 250ms while building turn metadata header"
2026-02-12 18:26:13.384 [error] [CodexMcpConnection] cli:} message="2026-02-12T09:26:13.384662Z ERROR codex_core::rollout::recorder: Falling back on rollout system\n2026-02-12T09:26:13.384689Z  WARN codex_core::state_db: state db record_discrepancy: list_threads_with_db_fallback, falling_back"
2026-02-12 18:26:13.436 [error] [CodexMcpConnection] cli:} message="2026-02-12T09:26:13.435934Z ERROR codex_core::rollout::recorder: Falling back on rollout system\n2026-02-12T09:26:13.436018Z  WARN codex_core::state_db: state db record_discrepancy: list_threads_with_db_fallback, falling_back"
2026-02-12 18:26:13.485 [warning] [CodexMcpConnection] cli:} message="codex_core::turn_metadata: timed out after 250ms while building turn metadata header"
2026-02-12 18:26:13.752 [error] [CodexMcpConnection] cli:} message="2026-02-12T09:26:13.751368Z ERROR codex_core::rollout::recorder: Falling back on rollout system\n2026-02-12T09:26:13.751448Z  WARN codex_core::state_db: state db record_discrepancy: list_threads_with_db_fallback, falling_back"
2026-02-12 18:26:15.292 [warning] [CodexMcpConnection] cli:} message="codex_protocol::openai_models: Model personality requested but model_messages is missing, falling back to base instructions. model=gpt-5.1-codex-mini personality=pragmatic"
2026-02-12 18:26:15.503 [error] [CodexMcpConnection] cli:} message="2026-02-12T09:26:15.502926Z ERROR codex_core::rollout::recorder: Falling back on rollout system\n2026-02-12T09:26:15.502950Z  WARN codex_core::state_db: state db record_discrepancy: list_threads_with_db_fallback, falling_back"
2026-02-12 18:26:15.816 [error] [CodexMcpConnection] cli:} message="2026-02-12T09:26:15.816302Z ERROR codex_core::rollout::recorder: Falling back on rollout system\n2026-02-12T09:26:15.816333Z  WARN codex_core::state_db: state db record_discrepancy: list_threads_with_db_fallback, falling_back"
2026-02-12 18:26:20.450 [error] [CodexMcpConnection] Codex process fatal error: {}, signal={}, exitCode={} templateArgs=["Codex process errored: Codex app-server process exited unexpectedly (code=1 (0x1), signal=null). Last CLI error: 2026-02-12T09:26:15.816302Z ERROR codex_core::rollout::recorder: Falling back on rollout system\n2026-02-12T09:26:15.816333Z  WARN codex_core::state_db: state db record_discrepancy: list_threads_with_db_fallback, falling_back",null,1]
2026-02-12 18:26:20.450 [error] [CodexWebviewProvider] Fatal error: {} templateArgs=["Codex process errored: Codex app-server process exited unexpectedly (code=1 (0x1), signal=null). Last CLI error: 2026-02-12T09:26:15.816302Z ERROR codex_core::rollout::recorder: Falling back on rollout system\n2026-02-12T09:26:15.816333Z  WARN codex_core::state_db: state db record_discrepancy: list_threads_with_db_fallback, falling_back"]
2026-02-12 18:26:21.307 [warning] [CodexMcpConnection] Attempted to send app-server message but stdin is destroyed message={"id":"auth:83f5c876-df66-4035-b89d-3924356c8dd0","method":"getAuthStatus","params":{"includeToken":true,"refreshToken":false}}
2026-02-12 18:26:21.307 [error] [CodexMcpConnection] Codex process fatal error: {}, signal={}, exitCode={} templateArgs=["Codex process errored: Codex process is not available",null,null]
2026-02-12 18:26:21.307 [error] [CodexWebviewProvider] Fatal error: {} templateArgs=["Codex process errored: Codex process is not available"]
2026-02-12 18:26:31.831 [warning] [CodexMcpConnection] Attempted to send app-server message but stdin is destroyed message={"id":"auth:45834b24-ca5d-4392-b202-bdbc25c2d940","method":"getAuthStatus","params":{"includeToken":true,"refreshToken":false}}
2026-02-12 18:26:31.832 [error] [CodexMcpConnection] Codex process fatal error: {}, signal={}, exitCode={} templateArgs=["Codex process errored: Codex process is not available",null,null]
2026-02-12 18:26:31.832 [error] [CodexWebviewProvider] Fatal error: {} templateArgs=["Codex process errored: Codex process is not available"]
2026-02-12 18:26:43.860 [warning] [CodexMcpConnection] Attempted to send app-server message but stdin is destroyed message={"id":"auth:01941f80-a37c-41b1-af67-846b8cb8ad84","method":"getAuthStatus","params":{"includeToken":true,"refreshToken":false}}
2026-02-12 18:26:43.860 [error] [CodexMcpConnection] Codex process fatal error: {}, signal={}, exitCode={} templateArgs=["Codex process errored: Codex process is not available",null,null]
2026-02-12 18:26:43.860 [error] [CodexWebviewProvider] Fatal error: {} templateArgs=["Codex process errored: Codex process is not available"]
2026-02-12 18:26:58.384 [warning] [CodexMcpConnection] Attempted to send app-server message but stdin is destroyed message={"id":"auth:4d775ed1-fb4e-49dc-b7e8-31f83d842a0d","method":"getAuthStatus","params":{"includeToken":true,"refreshToken":false}}
2026-02-12 18:26:58.385 [error] [CodexMcpConnection] Codex process fatal error: {}, signal={}, exitCode={} templateArgs=["Codex process errored: Codex process is not available",null,null]
2026-02-12 18:26:58.385 [error] [CodexWebviewProvider] Fatal error: {} templateArgs=["Codex process errored: Codex process is not available"]
2026-02-12 18:26:59.997 [warning] [CodexMcpConnection] Attempted to send app-server message but stdin is destroyed message={"id":"auth:afac76e8-bcc9-4d9b-b37e-f3cab91c106f","method":"getAuthStatus","params":{"includeToken":true,"refreshToken":false}}
2026-02-12 18:26:59.997 [error] [CodexMcpConnection] Codex process fatal error: {}, signal={}, exitCode={} templateArgs=["Codex process errored: Codex process is not available",null,null]
2026-02-12 18:26:59.997 [error] [CodexWebviewProvider] Fatal error: {} templateArgs=["Codex process errored: Codex process is not available"]
2026-02-12 18:27:00.023 [warning] [CodexMcpConnection] Attempted to send app-server message but stdin is destroyed message={"id":"auth:33abb559-126f-4e6b-97d3-ad8d832abb16","method":"getAuthStatus","params":{"includeToken":true,"refreshToken":false}}
2026-02-12 18:27:00.023 [error] [CodexMcpConnection] Codex process fatal error: {}, signal={}, exitCode={} templateArgs=["Codex process errored: Codex process is not available",null,null]
2026-02-12 18:27:00.023 [error] [CodexWebviewProvider] Fatal error: {} templateArgs=["Codex process errored: Codex process is not available"]
2026-02-12 18:31:45.627 [warning] [CodexMcpConnection] Attempted to send app-server message but stdin is destroyed message={"id":"auth:ee3f0e3c-090f-4ab0-b4a7-8b1519159894","method":"getAuthStatus","params":{"includeToken":true,"refreshToken":false}}
2026-02-12 18:31:45.627 [error] [CodexMcpConnection] Codex process fatal error: {}, signal={}, exitCode={} templateArgs=["Codex process errored: Codex process is not available",null,null]
2026-02-12 18:31:45.627 [error] [CodexWebviewProvider] Fatal error: {} templateArgs=["Codex process errored: Codex process is not available"]
---

What steps can reproduce the bug?

  1. Open Codex in IDE.
  2. Send a minimal prompt: "What time is it?".
  3. Codex crashes immediately when using tools.

What is the expected behavior?

no error.

Additional information

_No response_

View original on GitHub ↗

8 Comments

etraut-openai contributor · 5 months ago

You could try deleting (or temporarily renaming) your .codex directory. It's possible there's some corrupt state there.

Another idea is to install and run the Codex CLI manually from a terminal. The VS Code Extension is built upon the Codex CLI, and it appears that the Codex CLI process is crashing. By running it directly, we might get a better idea of why the crash is occuring.

tkysB · 5 months ago

Thanks for taking the time to look into this, I really appreciate it.

I already tried deleting the .codex directory as you suggested, but it didn’t resolve the issue. I can now access 5.3-codex via GitHub Copilot, but I’d still like the Codex extension itself to be fixed.

etraut-openai contributor · 5 months ago

I think this is the only report of a crash that we've received so far for this version of the extension, so it must be something that's fairly specific to your machine or configuration. I'll continue to look for other reports that might give us additional clues.

tkysB · 5 months ago

Thanks again, and I look forward to your update.

Master-cai · 5 months ago

Hi, I encounter similar issue:

Codex process errored: Codex app-server process exited unexpectedly (code=unknown, signal=SIGSEGV). Last CLI error: 2026-02-15T04:10:02.969491Z ERROR codex_core::rollout::list: state db missing rollout path for thread 019c59d0-2fa8-7e60-bf81-c72b86702217
2026-02-15T04:10:02.969512Z  WARN codex_core::state_db: state db record_discrepancy: find_thread_path_by_id_str_in_subdir, falling_back
2026-02-15T04:10:02.969491Z ERROR codex_core::rollout::list: state db missing rollout path for thread 019c59d0-2fa8-7e60-bf81-c72b86702217
2026-02-15T04:10:02.969512Z  WARN codex_core::state_db: state db record_discrepancy: find_thread_path_by_id_str_in_subdir, falling_back
Viomnz · 4 months ago

On Windows, Codex sometimes crashes when interacting with PowerShell or the shell snapshot system, which is not fully supported yet

sunwolf7 · 2 months ago

I am encountering a very similar issue on Windows 11 and spent several hours isolating the problem. I performed a full VS Code reinstall and extensive troubleshooting. One important finding is that Codex CLI works reliably while the VS Code extension and Codex app remain unstable. Detailed findings below.

Title: Codex VS Code extension and Codex app repeatedly crash/disconnect on Windows 11; Codex CLI works

Environment:

  • OS: Windows 11
  • IDE: Freshly reinstalled VS Code
  • Codex extension: latest version installed fresh
  • Codex CLI: installed and appears functional
  • Network: home network with Pi-hole and Tailscale normally installed; issue persists after extensive reset/testing, though disabling Tailscale briefly allowed one basic VS Code extension task to complete
  • Project tested: Tauri 2 / React / TypeScript / Rust / SQLite repo
  • Also tested: empty/minimal folder with only README.md

Problem:
The Codex VS Code extension and Codex app are unreliable to the point of being unusable. The extension frequently crashes with “Codex process errored: Codex process is not available.” The Codex app starts some tasks but never completes them. Codex CLI, however, can inspect the same repo successfully.

Observed errors:

  • Codex process errored: Codex process is not available
  • codex_app_server::transport: dropping message for disconnected connection: ConnectionId(0)
  • codex_core_plugins::manifest: ignoring interface.defaultPrompt: maximum of 3 prompts is supported path=C:\Users\sunwo\.codex\.tmp\plugins\plugins\openai-developers\.codex-plugin\plugin.json
  • codex_core::shell_snapshot: Failed to create shell snapshot for powershell: Shell snapshot not supported yet for PowerShell
  • codex_core::endpoint::responses_websocket: failed to connect to websocket: IO error: task was canceled, url: wss://chatgpt.com/backend-api/codex/responses
  • codex_core::tasks: task ... didn't complete gracefully after 100ms
  • Usage tab in Codex settings often fails to load unless reloaded multiple times.

Steps attempted:

  1. Restarted VS Code and reloaded the VS Code window many times.
  2. Switched Codex Windows sandbox from elevated to unelevated.
  3. Deleted C:\Users\_____\.codex\.tmp\plugins.
  4. Renamed/deleted the full C:\Users\_____\.codex directory.
  5. Fully uninstalled VS Code.
  6. Deleted:
  • %USERPROFILE%\.codex
  • %USERPROFILE%\.vscode\extensions
  • %APPDATA%\Code\User\globalStorage
  • %APPDATA%\Code
  • %APPDATA%\Code_shared
  • %USERPROFILE%\.vscode
  1. Reinstalled VS Code from a fresh installer.
  2. Installed only the Codex extension.
  3. Signed into Codex again.
  4. Accepted the sandbox setup prompt.
  5. Tested both the real repo and a minimal folder.
  6. Disabled Tailscale and bypassed Pi-hole DNS during some tests.
  7. Tried both long prompts and tiny prompts.
  8. Tested Codex CLI separately; CLI was able to inspect the Constellation repo.

Minimal reproduction:

  1. Fresh VS Code install.
  2. Install Codex extension.
  3. Sign in.
  4. Open a folder containing only README.md.
  5. Ask: “Please inspect this workspace and summarize the project structure. Do not edit files.”
  6. Extension crashes with Codex process errored: Codex process is not available, sometimes with plugin manifest or disconnected connection errors.

Expected behavior:
The VS Code extension and Codex app should complete basic tasks, especially in a one-file test project.

Actual behavior:
The VS Code extension repeatedly crashes or disconnects. The Codex app also fails to complete tasks. Codex CLI appears to work, which suggests the issue is specific to the VS Code extension/app runtime, app-server, plugin, sandbox, or websocket/session layer rather than the repo itself.

Impact:
This has consumed several hours of troubleshooting and makes the VS Code extension and app unusable for development. I need guidance on what logs to provide and whether this is a known current Windows extension/app issue.

Additional note:
This report was assembled with the help of ChatGPT after several hours of structured troubleshooting. The troubleshooting process included environment resets, extension reinstalls, sandbox configuration changes, DNS/network isolation tests, and comparison testing between the VS Code extension, Codex app, and Codex CLI.

geratorabora · 12 days ago

Same crash on Windows — adding a likely trigger and a workaround that actually works.

Env: extension openai.chatgpt 26.623.101652, Windows 11, model gpt-5.5. Same signature:
infinite "Thinking", existing sessions open as a blank editor tab, and the log shows the state_db
discrepancy → Codex process errored: Codex process is not available, with the app-server exiting
(code=4294967295 / 0xffffffff) and crash-looping. My exact line:
codex_rollout::state_db: state db discrepancy during read_repair_rollout_path: upsert_needed (fast path)

Likely trigger — concurrent app-servers on one CODEX_HOME. At the time of failure I had two
VS Code windows open with Codex active in both (two projects). Each window spawns its own
codex.exe, and both write the shared ~/.codex/state_5.sqlite. @tkysB — were you also running
Codex in more than one window/instance? This looks like a concurrent-write race on the state DB;
I don't see it mentioned above.

The DB is NOT corrupt — it's a logical/concurrency discrepancy. I backed up and inspected the
failing state_5.sqlite with SQLite:

  • PRAGMA integrity_check = ok, quick_check = ok, foreign_key_check = empty.
  • threads: every row has a rollout_path and every referenced rollout file exists on disk; no

duplicates, no empties; migrations + backfill complete.
So the file is structurally fine — read-repair hits a transient inconsistency and the app-server
panics instead of degrading.

Workaround that worked for me (reinitializing all of ~/.codex did NOT work for OP; this
narrower reset did, and it keeps history):

  1. Kill all codex.exe.
  2. Back up ~/.codex/state_5.sqlite*.
  3. Rename state_5.sqlite, -wal, -shm aside (don't delete).
  4. Reload window → the app-server rebuilds a clean state_5.sqlite and rescans sessions/ +

archived_sessions/. Conversation history is preserved (it lives in the rollout files, not the DB).

Suggested fixes: (1) read-repair should log+skip a bad entry, never abort the whole app-server;
(2) surface the fatal backend error in the UI instead of an infinite silent spinner; (3) an advisory
single-instance lock per CODEX_HOME so a second codex.exe can't race on the shared SQLite state.
Happy to share the intact state_5.sqlite and full Codex.log.