Environments

Every project has two environments - test and live - that are fully isolated from each other. They exist so you can develop, integrate, and load-test without touching production traffic.

What isolation means

The two environments share a project's name and region but nothing else:

A test key literally cannot resolve live resources - they don't exist in its view of the world.

The promotion workflow

The intended lifecycle for a change:

1. Develop in test - build against mrnd_test_ keys with no risk to production.
2. Verify - webhook deliveries, event payloads, and sync behavior all work identically in test.
3. Promote configuration - recreate webhook endpoints and settings in live (manually, or via the Environments API / Terraform).
4. Switch credentials - deploy mrnd_live_ keys and flip your config flag.

# test
curl https://api.meridian.dev/v2/projects \
  -H "Authorization: Bearer $MERIDIAN_TEST_KEY"

# live - identical path, different key
curl https://api.meridian.dev/v2/projects \
  -H "Authorization: Bearer $MERIDIAN_LIVE_KEY"

The environment is selected entirely by the API key - there is no ?env= parameter and no way to mix them in one request.

Test environment rules

• Test data may be pruned after 90 days of inactivity.
• Outbound calls from test webhooks are allowed to any URL, including localhost tunnels like localhost:3000 or ngrok.
• Test event payloads are structurally identical to live ones - only the environment field and volume differ.

Managing environments via API

Environments themselves are resources - useful for infrastructure-as-code setups:

curl https://api.meridian.dev/v2/projects/prj_9kLm2xVb7nQe/environments \
  -H "Authorization: Bearer $MERIDIAN_API_KEY"

{
  "object": "list",
  "data": [
    {
      "id": "env_2hF8kLm4xQ",
      "object": "environment",
      "name": "test",
      "status": "active",
      "event_retention_days": 7
    },
    {
      "id": "env_7dR1mPx9vW",
      "object": "environment",
      "name": "live",
      "status": "active",
      "event_retention_days": 90
    }
  ]
}

Common mistakes

• Using a test key in production config - calls succeed but hit an empty environment; resources "missing" in production are usually just in the other environment.
• Registering production webhooks in test - deliveries silently go nowhere. Check the environment badge in the dashboard before creating endpoints.
• Hardcoding keys per environment - inject the key via config so promoting to live is a one-line change.