[Bug]: configure GitHub copilot as provider

[ilteoood]

Affected component

provider, onboard

Severity

S2 - workflow degraded (workaround exists: manual config.toml edit)

Current behavior

The zeroclaw onboard command does not present GitHub Copilot as a selectable provider. Copilot is a supported provider with existing implementation, but it is absent from the onboarding wizard's provider list. Users who want to use Copilot must manually edit config.toml:

default_provider = "copilot"
default_model = "gpt-5-mini"

Then run zeroclaw agent, which will trigger a browser-based Copilot login flow on the first prompt. There is no documentation for this workaround.

Expected behavior

zeroclaw onboard should include GitHub Copilot in the provider selection list and guide the user through the OAuth device-flow login, the same way other providers prompt for credentials.

Steps to reproduce

zeroclaw onboard
# GitHub Copilot does not appear in the provider list

Impact

All users who want to use GitHub Copilot as their provider cannot do so through the standard setup path.

Logs / stack traces

N/A

ZeroClaw version

0.6.5

Rust version

No response

Operating system

Docker

Regression?

No — Copilot was never added to the onboarding wizard.

Pre-flight checks

• I reproduced this on the latest master branch or latest release.
• I redacted secrets, tokens, and personal data from all submitted content.

Fix

PR #5321 adds Copilot to the onboarding command. This issue will close when that PR merges.

[jamieclaw]

I found this issue quite annoying as well. The copilot is not in the onboard command.

But you can manually edit the config.toml file:

default_provider = "copilot"
default_model = "gpt-5-mini"

Then run zeroclaw agent and type anything, it will prompt for the copilot login:

After authenticate success, it will connect to copilot's llm, then, just do zeroclaw gateway restart to restart the gateway.

[zafranf]

==== EDITED
it works now, maybe its because one of api_key in config doesnt in the right place
=====.

I found this issue quite annoying as well. The copilot is not in the onboard command.

But you can manually edit the config.toml file:

default_provider = "copilot"
default_model = "gpt-5-mini"

Then run zeroclaw agent and type anything, it will prompt for the copilot login:

After authenticate success, it will connect to copilot's llm, then, just do `zeroclaw gateway restart` to restart the gateway.

hi @jamieclaw, i followed you but didnt get the same result as yours, do you have any clue?

[theonlyhennygod]

Closing — addressed by an existing PR.