Some API endpoints return 500 Internal Server Error since Aug 7, 2025

[YasinC2]

Starting around August 7, 2025, running wgcf register no longer works. It consistently fails with a 500 Internal Server Error from the Cloudflare API, both locally and in GitHub Actions. This issue was not present before and the same setup was working for weeks.

Steps to Reproduce:

1. Download the latest wgcf (tested on v2.2.27).

2. Run:

   wgcf register --accept-tos

3. Observe output:

   2025/08/09 17:15:24 Using config file: wgcf-account.toml
   2025/08/09 17:15:26 500 Internal Server Error
   

Expected Behavior:
wgcf should register successfully and generate wgcf-account.toml.

Actual Behavior:
Fails with:

500 Internal Server Error

and exits with code 1.

• Local test (Windows 11): Even though the 500 error appears, wgcf-account.toml is still created.

• GitHub Actions (Ubuntu 24.04): The same command fails with a non-zero exit status, which stops the workflow before the wgcf-account.toml file can be used in later steps.

Environment:

• wgcf version: v2.2.27
• OS tested: Windows 11 Pro 64-bit, Ubuntu 24.04 GitHub Actions runner
• Python version: 3.x
• Network: Normal residential connection + GitHub-hosted runners

Additional Notes:

• This happens both locally and in GitHub Actions.
• No changes were made to the script or environment before the failure started.
• Possibly caused by Cloudflare changing or breaking the WARP registration API.

Could you confirm if this is a known Cloudflare API change and if a patch or workaround is available?

[wilnguyen]

Same behavior on MacOs Sequoia and Tahoe 26 Public Beta.
The Wrap client from Cloudflare is still fine

[SHJordan]

Same issue, subscribing to the thread.

[ViRb3]

Yes, I can confirm the same here. I will investigate.

[JeelsBoobz]

Yes, I can confirm the same here. I will investigate.

Thanks sir

[diepnt90]

Same here

[tempppabx1]

Yes, I can confirm the same here. I will investigate.

I suspect the error is caused by the device name configuration and the active operation in the /reg endpoint. It's noteworthy that skipping these steps still results in a functional configuration file. Cloudflare may have recently changed or deprecated some of this functionality, but this requires further investigation.

[umatrix81]

Even if register fails, you still can use generate to get valid config.

[ViRb3]

Did some tests, this appears to be a bug in Cloudflare's Android API as their official app also results in "Unknown error occurred" when you try to rename a device. In either case, I will release an update to ignore this error in wgcf as it's not critical.

[ViRb3]

Seems like more APIs are broken and return 500 as well. This will be a larger update than I thought.

[ViRb3]

This should be fixed in https://github.com/ViRb3/wgcf/releases/tag/v2.2.28

To summarize, UpdateSourceBoundDeviceActive and SetDeviceName always return error 500, but despite that they still perform their action as expected. This also happens on the latest version of 1.1.1.1 Android app, so it must be a server-side issue. I put some workarounds in the code and wgcf appears to work again. This is not an ideal fix though, so will keep the issue open until we fix it properly, either on our side or Cloudflare's side.

[Manouchehri]

this appears to be a bug in Cloudflare's Android API as their official app also results in "Unknown error occurred" when you try to rename a device.

It happens on the official iOS app as well.

[ViRb3]

Discovered another bugged endpoint as part of 03c9c1f - deactivating devices will return 500 and remove them instead.