Create an App and Add Your OpenAI Token
In this tutorial you will:
- Log in to your Qelos instance
- Add an OpenAI connection (where the API key lives)
- Pull the configuration locally with the CLI
Estimated time: 10 minutes
Prerequisites
- A Qelos instance — either sign up at app.qelos.io for a managed cloud instance, or run it locally by following the Installation Guide.
- An OpenAI account with an active API key. You can create one at platform.openai.com.
Step 1 — Log In to Your Qelos Admin
Open your browser and navigate to your Qelos instance (e.g. http://localhost:3000). Log in with your admin credentials.
Default development credentials:
test@test.com/admin
Step 2 — Add an OpenAI Connection
In Qelos, a connection stores credentials for an external provider (like OpenAI). Agents (integrations) are built on top of connections — they add the model, system prompt, and behaviour, but the API key itself lives in the connection.
- In the left sidebar, navigate to Integrations → Connections.
- Find OpenAI in the list of available providers and click it.
- Click Create to add a new OpenAI connection.
- Paste your OpenAI API key into the API Key field.
- Click Save.
Security note: Your API key is stored encrypted in the Qelos database and is never exposed to frontend clients.
Step 3 — Pull Your Config Locally (Optional)
If you plan to use the CLI or SDK, pull the full workspace configuration to your local project. This downloads connections, integrations, blueprints, and more into your working directory.
Configure authentication via .env
Create a .env file in your project root with your Qelos instance URL and credentials:
# Required
QELOS_URL=https://your-instance.qelos.io # or http://localhost:3000 for local
# Option A — username & password
QELOS_USERNAME=your@email.com
QELOS_PASSWORD=yourpassword
# Option B — API token
QELOS_API_TOKEN=your-api-tokenPull everything
qelos pull all ./You will see local files for each resource — connections, integrations, blueprints, and more. The qelos agent command uses these files to resolve integrations by name.
What's Next?
Now that your OpenAI connection is in place you can:
- Configure your first AI agent — create an integration (agent) that uses this connection, give it a persona, and start chatting with it.
- AI Operations SDK reference — call the AI from your own code.
- CLI Agent command — chat with an integration directly from the terminal.