Runbooks, not dashboards
A wall of graphs tells you a number went up. A runbook tells you what you decided to care about, and what to do when it breaks. We built Vigil around the second idea.
Dashboards are where intent goes to die. Someone sets one up, it’s glorious for a week, and then it slowly fills with panels nobody remembers adding and thresholds nobody trusts. The graph is a record of what happened, not a statement of what you meant to guarantee.
A runbook is a promise
In Vigil a runbook is a named, scheduled set of rules attached to one unit. “Checkout stays up, TLS stays valid, the login API answers under 800ms.” Each rule is a typed check from a catalogue with explicit inputs — not a free-form query you have to get right at 3am.
Because the rule is typed, Vigil validates it when you save it, runs it on a schedule, and keeps its history. The runbook becomes the single place that answers “what are we actually promising for this client?”
Graphs are still there when you need them
This isn’t anti-metric. Every check keeps its timing and its evidence, so you can still see the latency creep before it became a breach. The difference is that the graph hangs off a decision you made on purpose, instead of standing in for one.