FAQs & Troubleshooting
Provider FAQs & troubleshooting
I saved an API key but can't see it anymore. Is it lost?
No — this is by design. Keys are stored encrypted and write-only: after saving, the key is never displayed again, not even to you. Adri only shows whether a key is Configured or Not set. To rotate it, paste a new key over the old one and click Save key.
Test connection failed
Check these in order:
- Is the key correct? Paste it again (a stray space or truncated paste is the most common cause) and click Save key, then test again.
- Is the key active? Open your provider's console and confirm the key is not revoked, expired, or out of credit.
- Fix, save, and click Test connection again. Expect OK — N model(s).
The model picker says "No providers configured"
Your project has no enabled providers. Either none was added yet — follow Add a provider — or they are all switched off on the Providers settings page.
A provider doesn't appear in the picker
- Check its switch on the Providers settings page: it must read enabled.
- The picker lists providers per project — make sure you added the provider to the same project as the conversation.
A provider shows "No models available"
The model list comes from a built-in catalog plus live discovery against your endpoint. If it is empty:
- Add and test the API key first — discovery needs a working connection.
- Re-run Test connection: OK — 0 model(s) means the endpoint answered but offered no models.
A model shows "(unavailable)"
The conversation was pinned to a provider that has since been removed. Nothing is lost — open the model picker and pick a new model, or Adri AI for the project default. In VS Code the picker opens automatically when this happens.
Who can add or change providers?
Any member of the project can manage its providers from Settings → Project Settings → Providers. Providers are shared project-wide.
Does switching models change my conversation history?
No. A switch takes effect on the conversation's next message. Existing messages are never re-generated or rewritten.
What does the "Adri AI" entry actually run?
Adri AI means "inherit the project default": the project's default provider (the one with the default badge) and its default model. If your project has no providers, it is the Adri-managed model — this is how every project starts.
Who pays for the tokens?
Conversations on your provider call the model with your API key, so usage is billed to your provider account (Anthropic, OpenAI, …). Conversations on Adri AI without a default provider configured run on Adri-managed capacity.
Still stuck?
Contact us — pick the channel that fits your need.