Gloo AI Python Quickstart (pipenv)

This minimal example was built in response to the Slack thread in #support-gloo-ai (https://servant-io.slack.com/archives/C08QPAK2QAX/p1764093157738929) asking for a working Python snippet after a developer hit 403 forbidden when calling the quickstart API.

What this does

• Loads GLOO_AI_CLIENT_ID and GLOO_AI_CLIENT_SECRET from a local .env.
• Requests an OAuth token with the api/access scope.
• Calls the chat completions endpoint using meta.llama3-70b-instruct-v1:0.
• Prints the token expiration and the JSON completion response.

For the full walkthrough, see the official quickstart docs: https://docs.gloo.com/getting-started/quickstart-developers.

Setup

1. Ensure Python 3.11 is available.
2. Create .env (already gitignored) with:

   GLOO_AI_CLIENT_ID=...
   GLOO_AI_CLIENT_SECRET=...
   

   The provided demo values are already placed in .env for convenience; replace if you have org-specific credentials.
3. Install deps and lock via pipenv:

   pipenv install --python 3.11

Run

pipenv run gloo-chat

Format

pipenv run format

Uses black via the Pipfile script to format the repository.

Notes / troubleshooting

• A 403 usually means the token lacks the right scope or the Authorization header is malformed. This example uses grant_type=client_credentials with scope=api/access and sets Authorization: Bearer <token>, matching the published quickstart.
• If you rotate credentials, update .env and rerun.