GET STARTED

Connect models to your tools.

Enter your delivered endpoint, key and model ID in the custom provider settings. Check whether your client appends paths automatically.

Read credentials from environment variables. Public examples use placeholders; actual values arrive by private delivery email.

Share your tool name and version before purchasing. Confirm compatibility first, then follow the instructions provided.

RUN7 / FIELD GUIDE

A practical setup checklist.

Back up the client configuration before adding a custom provider. Avoid replacing a working account until your new connection passes a short test.

Copy the endpoint exactly, use your own dedicated key and select an exact model ID. A display name is not always a valid request ID.

Send a short text prompt. Confirm a response, then check the model and charge in the private usage portal before running a long task.

RUN7 / FIELD GUIDE

Find custom provider settings and confirm OpenAI compatibility. Check whether the client adds /v1 automatically; a duplicated path can cause a 404.

Tool calling and native protocols matter. Share the tool name and version before buying; do not assume every model supports every agent feature.

Read secrets from environment variables, set request timeouts and limit concurrency. Never bundle a customer key in a public frontend.