Docs / GitHub Copilot provider
Use GitHub Copilot as a model provider
If your organisation already pays for GitHub Copilot, you can run Renly chat turns on your own Copilot subscription. Usage bills your GitHub AI Credits instead of your Renly quota.
How it works
Renly integrates the official GitHub Copilot SDK. You connect a GitHub token for your own account, Renly verifies it against Copilot, and from then on Copilot's models appear in your chat model picker with a (Copilot) label. Turns on those models run through your Copilot seat.
Connect
- Make sure your GitHub account has an active Copilot seat (any paid Copilot plan, including a seat from your organisation's Copilot Business or Enterprise subscription).
- Create a GitHub token for your account. A fine-grained personal access token works.
- In Renly, open Connections, find the GitHub Copilot card, paste the token, and choose Connect. Renly verifies the token against Copilot before storing it, encrypts it at rest, and never displays it again.
- Open any chat and pick a Copilot model from the model picker.
What to expect today
| Works now | Coming next |
|---|---|
| Chat turns on Copilot models, streamed live, with the conversation as context | Workspace tools and connectors on Copilot-billed turns, governed by the same approval flow |
| Your full Copilot model list, fetched from your subscription | Organisation-level Copilot policies |
| Billing to your GitHub AI Credits |
Copilot-billed turns currently answer from the conversation alone. When a turn needs files, connectors, or other tools, switch to a Renly model for that turn; the chat itself keeps working exactly the same.
Security
- Your token is verified before it is stored, encrypted at rest, and removable at any time with Disconnect.
- Copilot sessions run fully contained: no shell, file, or network tools are exposed to them.
- Connecting and disconnecting are recorded in your organisation's audit trail.