Quickstart

HALO gives one project-scoped key access to provider-native model APIs, long-term Memory, Authentication, service identity, and usage controls. This quickstart sends a first OpenAI-compatible model request.

#Before you begin

You need a HALO account and an active project. Model calls consume the account balance assigned to the project owner.

#1. Create a project

Open the HALO dashboard, choose Projects, and create a project for the application or environment you are integrating.

1. 1
   Name the boundary

   Use separate projects for production, staging, and unrelated OEM products. Memory and Authentication are isolated by project.

2. 2
   Open the project

   The selected project becomes the dashboard context for keys, models, Memory, history, and Authentication.

#2. Create an API key

Select API Keys, create a client key, and copy it once. Store it in your server secret manager as HALO_API_KEY.

HALO_API_KEY=sk-your-project-client-key

#3. Make a request

HALO keeps each vendor's familiar wire protocol. Point an existing OpenAI client at the HALO base URL and keep the rest of your request structure.

curl https://api.agihalo.com/openai/v1/chat/completions \
  -H "Authorization: Bearer $HALO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-5-mini",
    "messages": [
      { "role": "user", "content": "Reply with one word: ready" }
    ]
  }'

#Next steps