How should I troubleshoot a model connection error in my agent?
Trace failures through the agent configuration, model connection, credential access, and provider service.
Maintained by the 路墨 Lumoswitch product team
Confirm the active configuration path and model ID first, then check keys, provider access, limits, and request records.
What should I check in the agent first?
Confirm that the agent was reopened and that you launched the installation and user directory you configured. Check the active model ID, protocol, and connection path. Project configuration or environment variables may override user-level settings. For local LiteLLM, make sure Lumoswitch is running; for an online connection, confirm your account state.
How do I check the credential and provider next?
Verify that the key is active, the provider account can use the selected model, and the endpoint and protocol match. For a local connection, check that the local service is running. Check provider-side balance, rate limits, and service status. Do not test by pasting a key into a command, screenshot, or chat.
What evidence should I send to support?
Record the failure time, agent, model ID, connection type, a redacted error code, and app version. For platform calls, inspect the matching request status, model, usage, or routing record. A local history log from an external agent is not the same data source as a Lumoswitch platform bill. Remove keys, prompt contents, and personal information before filing a support request.