OpenAI Connector

Learn how to integrate OpenAI's models into your graph

OpenAI provides AI models like GPT-4 for text generation, completion, and processing tasks. The prebuilt OpenAI Connector provides a GraphQL interface for OpenAI's models and services, allowing you to integrate OpenAI's capabilities into your federated graph.

Connector capabilities

This Connector can incorporate the following OpenAI APIs into your graph:

Prerequisites

To use the Connector, you need an OpenAI API key.

Try out the Connector

Use the Apollo Sandbox below to test out the Connector. This is the same interface you use locally to test your queries.

To use this sandbox you either need to add your OpenAI key as a header or copy and paste placeholder values.

Add your OpenAI API key
  1. In the sandbox, below the Operations panel, open the Headers tab then click Add a new header.
  2. Add a new header with the name Bearer and your OpenAI token value.
  3. Select and run a query from the Documentation tab to test.
Use placeholder values
  1. Copy and paste the following into the Operations panel in Sandbox:
    GraphQL
      mutation GenerateStory($model: ID!, $about: [MessageInput]!) {
  createChatCompletion(model: $model, messages: $about) {
    id
    model
    messages {
      content
      role
    }
    usage {
      totalTokens
      promptTokens
      completionTokens
    }
    createdAt
  }
}
  2. Copy and paste the following into the Variables panel below Operations:
    JSON
    {
  "about": [
    {
      "role": "USER",
      "content": "The story about Apollo"
    }
  ],
  "model": "gpt-4o-mini"
}
  3. Run the operation.

Getting started

  1. If you haven't already, create a new graph in GraphOS.

  2. Copy the supergraph.yaml and router.yaml files from the Connector to replace the files created by running rover init.

  3. Set your OpenAI API key as an environment variable in your terminal:

    terminal
    export OPENAI_API_KEY=....

  4. Run rover dev to start the local development session:

    terminal
    APOLLO_KEY=service:My-Graph-s1ff1u:•••••••••••••••••••••• \
  APOLLO_GRAPH_REF=My-Graph-s1ff1u@main \
  rover dev --supergraph-config supergraph.yaml --router-config router.yaml

You're all set! Open http://localhost:4000 to query your graph using Apollo Sandbox.

Adding to an existing graph in GraphOS

To add this Connector to an existing graph, publish the schema openai.graphql file to your graph ref using rover subgraph publish:

terminal
APOLLO_KEY=service:My-Graph-s1ff1u:•••••••••••••••••••••• \
  rover subgraph publish My-Graph-s1ff1u@main --name openai --schema openai.graphql --routing-url http://openai

Setup VS Code task runner

Once you've set up the VS Code task runner, you can execute the Tasks: Run Task command in VS Code to run the rover dev task. Edit your .vscode/settings.json to include the following OpenAI-specific key:

terminal
{
  "terminal.integrated.profiles.osx": {
    "graphos": {
      "path": "zsh",
      "args": ["-l"],
      "env": {
        "OPENAI_API_KEY": "",
        ...
      }
    }
  },
  "terminal.integrated.defaultProfile.osx": "graphos"
}

Alternatively, you can open a new terminal window in VS Code with the graphos profile, then run rover dev --supergraph-config supergraph.yaml --router-config router.yaml.

Contributing

The Connectors Community welcomes contributions to this Connector or to expand the library. For instructions on how to contribute, see the contributing guide.

Modules to contribute

The following features can be added to this Connector:

  • Support for all OpenAI models and endpoints

  • Stream response handling

  • Function calling support

  • Image generation and analysis

See the OpenAI API Documentation for other features to implement.
Feedback

Edit in VSCode

Ask Community