Problem
Nothing in the rules says an environment must run on infrastructure the org owns. Contributors weigh a shared third-party instance as a reasonable tradeoff for speed, so a test that fails tells us nothing (our code, their config, or their quota), and an incident lands in an account we can't see into or shut off. The MVP launch goal hit this directly on Curia-Regis/app#139.
Importance: High - unreliable verification blocks every notification and audit fix, and an incident on a foreign account has no containment.
Urgency: High - the decision is live on the MVP queue right now.
Priority: P1
Solution
Add a DEV rule: every environment runs on accounts the org owns, with separate credentials per environment held as repo secrets. We own the integration, the keys and the access. Reviewers reject a Solution that leans on a third-party controlled or gated instance rather than weighing it as a tradeoff.
References
Problem
Nothing in the rules says an environment must run on infrastructure the org owns. Contributors weigh a shared third-party instance as a reasonable tradeoff for speed, so a test that fails tells us nothing (our code, their config, or their quota), and an incident lands in an account we can't see into or shut off. The MVP launch goal hit this directly on Curia-Regis/app#139.
Importance: High - unreliable verification blocks every notification and audit fix, and an incident on a foreign account has no containment.
Urgency: High - the decision is live on the MVP queue right now.
Priority: P1
Solution
Add a DEV rule: every environment runs on accounts the org owns, with separate credentials per environment held as repo secrets. We own the integration, the keys and the access. Reviewers reject a Solution that leans on a third-party controlled or gated instance rather than weighing it as a tradeoff.
References