NodeFox

Lesson 12 · Intermediate

Connect a model provider

Configure model access without exposing keys in workflows or recordings.

Written guide · 3 min read · video not yet produced

Before you start

A signed-in NodeFox workspace. Use synthetic data and prepare outputs before enabling external writes.

  1. 01Prepare a restricted key

    Create a provider key using your organization’s policy and spending controls. Use a development project for learning. Never show the key in a video, source repository, prompt, screenshot or shared fixture.

  2. 02Configure Integrations

    Open the application’s Integrations view and choose the supported provider. Enter the key in the intended credential field. Select a model that the account can actually use; a model listed in a guide may not be available to your key.

  3. 03Test a small request

    Use one short, non-sensitive prompt in a Conversation node. Inspect the returned output and error message. Keep the test small enough to distinguish configuration problems from large-payload or rate-limit issues.

  4. 04Diagnose precisely

    Authentication errors usually require checking the key or account. Model-not-found errors require checking model access. Rate limits require reducing concurrency or waiting according to the provider’s guidance. Do not solve these problems by embedding credentials in code.

If the result is wrong

Browser-local orchestration does not mean prompts remain on the device when an external AI provider is invoked.