Skip to Content
Model GuideIntegrationsGeneric setup and troubleshooting

General setup and troubleshooting

For tools without a dedicated guide, first check that they allow a custom API address, credentials and model ID. A claim of “OpenAI compatibility” does not establish support for every client feature.

Host, base URL and endpoint

  • Host: the scheme and domain, such as https://api.fenno.ai.
  • Base URL: the prefix to which a client appends operation paths. The OpenAI SDK uses https://api.fenno.ai/v1.
  • Full endpoint: includes an operation, such as https://api.fenno.ai/v1/chat/completions.
ProtocolTypical base URLFinal request pathAuthentication
OpenAI Chat Completionshttps://api.fenno.ai/v1/v1/chat/completionsAuthorization: Bearer
OpenAI Responseshttps://api.fenno.ai/v1/v1/responsesAuthorization: Bearer
Anthropic Messageshttps://api.fenno.ai/v1/messagesx-api-key; follow the dedicated guide for tool-specific authentication
Geminihttps://api.fenno.ai/v1beta/models/{model}:generateContentx-goog-api-key

These base URLs assume the client appends the corresponding path. If it asks for a full endpoint, enter the complete request URL instead. Do not duplicate the version segment. See authentication.

Setup checklist

  1. Choose a custom provider with an explicit protocol type.
  2. Enter the API address and Fenno key. The key is not a model ID.
  3. Copy an exact model ID from the console or /v1/models; do not invent a provider prefix.
  4. Save and enable the provider, then select it in the current conversation.
  5. Send a minimal text request and check the corresponding Fenno usage record.
  6. Enable streaming, images and tools one at a time.

Isolate account and URL problems

This request checks authentication and model discovery on the OpenAI-compatible endpoint. It does not test every model capability:

export FENNO_API_KEY="<YOUR_FENNO_API_KEY>" curl --fail-with-body https://api.fenno.ai/v1/models \ -H "Authorization: Bearer $FENNO_API_KEY"
SymptomCheck first
401 / 403Key validity, account permissions and conflicting older credentials
404Duplicated path segments, exact model ID and protocol match
400Remove extra parameters and test the smallest supported request
429 / insufficient quotaRead the error and check quota; follow errors and retries
TimeoutNetwork, proxies, certificates and client timeout; avoid duplicating long-running tasks
Replies without Fenno usageThe conversation may still use a default provider; check the account, time and route
Text works but tools failTest tool arguments, model capability and tool-result messages separately

Security and support

Do not put API keys in public repositories, screenshots or frontend code. Never post complete environment variables, authentication headers or unredacted logs in public issues.

Include the tool and version, OS, protocol, exact model ID, request time and timezone, HTTP status, and any returned request ID in a support report. Remove keys, private prompts and project content before sharing a minimal reproduction.