VS Code extension repeatedly fetches connector logos, causing "Server overloaded; retry later" errors

What version of the IDE extension are you using?

2.1.119

What subscription do you have?

Pro

Which IDE are you using?

VS Code

What platform is your computer?

Linux 6.8.0-107-generic x86_64 x86_64

What issue are you seeing?

The Codex VS Code extension repeatedly logs connector logo fetch failures.

Example error:

2026-04-24 22:00:27.944 [error] Error fetching error={"code":-32001,"message":"Server overloaded; retry later."} url=/aip/connectors/asdk_app_69956675f2848191b08e0400f2695572/logo?theme=dark

In one VS Code session, my Codex log contained around 1148 /aip/connectors/.../logo fetch errors.

I also saw nearby app-server errors:

Request failed conversationId=none error={"code":-32001,"message":"Server overloaded; retry later."} method=skills/list
Request failed conversationId=<redacted> error={"code":-32001,"message":"Server overloaded; retry later."} method=thread/read
Request failed conversationId=<redacted> error={"code":-32001,"message":"Server overloaded; retry later."} method=thread/resume

What steps can reproduce the bug?

1. Open VS Code with the Codex extension installed.
2. Sign in to Codex/ChatGPT.
3. Open the Codex panel.
4. Let the extension load apps/connectors.
5. Check the Codex extension log.

On my machine, the log was here:

~/.config/Code/logs/20260424T215827/window1/exthost/openai.chatgpt/Codex.log
The log repeatedly contained errors like:

text

Error fetching error={"code":-32001,"message":"Server overloaded; retry later."} url=/aip/connectors/<connector-id>/logo?theme=dark
Error fetching error={"code":-32001,"message":"Server overloaded; retry later."} url=/aip/connectors/<connector-id>/logo?theme=light
I do not have a small code snippet because this is an IDE extension / app-server behavior, not repository code.

What is the expected behavior?

The extension should not flood the log with repeated connector logo fetch failures.

Expected behavior:

• Connector logos should be cached, throttled, or retried quietly.
• A logo fetch failure should not produce hundreds or thousands of repeated error lines.
• Logo fetch failures should not affect unrelated operations such as thread/read, thread/resume, or skills/list.
• If the server is overloaded, the extension should back off instead of retrying many logo requests.

Additional information

I manually tested one of the same logo requests using the local Codex auth token, without printing the token.

Authenticated backend request:

GET https://chatgpt.com/backend-api/aip/connectors/asdk_app_69956675f2848191b08e0400f2695572/logo?theme=dark

Result:

status: 200 OK
content-type: image/png
content-length: 10205

So the logo itself was reachable through the authenticated backend API. The issue seems to be the extension/app-server retry behavior or request burst behavior.

Environment:

IDE extension version: 26.422.21459
IDE: VS Code
Platform: Linux 6.8.0-107-generic x86_64 x86_64

I also saw separate telemetry errors like this, but I think they are unrelated:

Error fetching httpStatus=403 statusText=Forbidden url=https://chatgpt.com/ces/v1/rgstr?<redacted>