Configure a direct connection
- Open Settings → Models and configure the provider API key or Foundry connection you intend to use.
- Save the connection, then open Code Harnesses → Rudr Code.
- Enable Rudr Code in the harness visibility controls if needed and review its model/default choices.
- Choose Save settings, open a new Code session, and select Rudr Code and an available model in the composer.
- Check permission mode and the working folder before sending a request.
There is no external CLI to install for this harness and no Claude, ChatGPT, or other subscription-login step that converts those allowances into Rudr Code API access.
Work within pre-alpha limits
Available models depend on the configured connection. A provider key’s presence does not prove the key is valid, its account can use the selected model, or a Foundry deployment is correctly mapped. Inspect the actual readiness and error states.
Rudr Code exposes tools and permission controls, but pre-alpha availability should not be interpreted as production qualification of every feature. Use a reviewable task in a folder whose changes you can inspect. Check the Files and Git panels after any editing work, and preserve the exact error if a run stops unexpectedly.
Recover configuration problems
If no model is available, return to Models and verify the connection and deployment selection. If a tool is missing, inspect Tools for the installed source and selected harness. Reinstalling another provider’s CLI does not repair Rudr Code’s direct connection.
See API connections, permissions, and availability before relying on optional features.