Connic Composer SDK

Environment Variables

Learn how to use environment variables to securely store API keys, database credentials, and configuration values that your agents can access at runtime.

What are Environment Variables?

Securely store secrets and configuration outside your code

Environment variables allow you to store sensitive data like API keys, database credentials, and configuration values separately from your code. They are secure (never in git),environment-specific (different values per environment), andeasy to update without code changes.

How Variables Are Injected

When you deploy your agents, Connic automatically injects all environment variables configured for that environment. Access them using standard Python:

python

import os

api_key = os.environ.get("MY_API_KEY")
log_level = os.environ.get("LOG_LEVEL", "INFO")

Variables are only applied on deployment. You must redeploy for changes to take effect.

Using Variables in Tools

python

# tools/api_client.py
import os
import httpx

async def call_external_api(query: str) -> dict:
    """Call an external API using credentials from environment variables."""
    api_key = os.environ.get("EXTERNAL_API_KEY")
    api_url = os.environ.get("EXTERNAL_API_URL", "https://api.example.com")

Using Variables in Middleware

python

# middleware/auth.py
import os
from connic import StopProcessing

async def before(payload: dict, context: dict) -> dict:
    """Validate requests using an internal auth service."""
    auth_url = os.environ.get("AUTH_SERVICE_URL")
    auth_secret = os.environ.get("AUTH_SERVICE_SECRET")
    
    if not auth_url or not auth_secret:
        raise StopProcessing("Auth service not configured")
    
    # ... validation logic using auth_url and auth_secret
    return payload

Environment-Specific Values

python

# Your code stays the same - values differ per environment
import os

database_url = os.environ.get("DATABASE_URL")  # Different per environment
api_key = os.environ.get("API_KEY")            # sk_test_xxx in staging, sk_live_xxx in prod
log_level = os.environ.get("LOG_LEVEL", "INFO")  # DEBUG in staging, WARNING in prod

Configure different values per environment in the dashboard. Your code stays the same.

Sensitive Variables

Mark variables as sensitive to hide them in the dashboard (shown as ••••••••) and mask them in logs. Always mark these as sensitive: API keys, passwords, database credentials, and tokens.

Best Practices

Use uppercase with underscores: DATABASE_URL not databaseUrl
Provide defaults for optional config: os.environ.get("LOG_LEVEL", "INFO")
Validate required variables early: Check at startup, not when first used
Use different values per environment: Test keys for staging, live keys for production

Important Notes

Redeploy required: Changes only take effect after redeployment.Don't commit secrets: Never put API keys in your repository.

Managing Variables in the Dashboard

Go to your project → Settings → Variables. Select the environment, add variables with the form, and redeploy to apply changes.

Write Custom Tools

Use variables in your tool code

Middleware

Use variables in middleware hooks

MCP Integration

Authenticate with MCP servers

Agent Configuration

Configure your agents