Proof Kit / Examples
Walkable examples that show Cassian Gate producing authoritative PASS and authoritative FAIL outcomes against real declared topologies. Each entry pairs a passing variant (the gate confirms the declared proof is satisfied) with a failing variant (the gate catches a real validation problem on the same bounded topology).
First-Run Proof
The official first-run adoption proof family. Failure-first path: run the failing variant first to see Cassian Gate catch a real validation problem, then run the passing variant to see the same proof succeed when the policy is correct.
Recipe — Can host A reach host B?
A bounded service-reachability check: host A reaches host B on a declared allowed TCP port. The next-step bridge after the first-run proof.
Recipe — Does this firewall block or allow a port?
Two bounded policy outcomes in one small example: TCP/8443 is allowed, TCP/2222 is blocked. The failing variant declares the wrong expectation for the actual policy.
Recipe — Does validation fail when a link drops?
A bounded deterministic failure-choreography example. A directly connected reachability topology with a deliberate interface bounce scenario, run only through the normal authoritative gate path.
Recipe — What changed between baseline and change?
Uses cassian test --two-run to validate a proposed change against a known-good baseline. Two clean-state authoritative gate executions plus an advisory comparison.json artifact recording what differs between the runs.
Recipe — How do I validate an IaC change?
Uses cassian adapt terraform to convert a terraform show -json plan into an advisory adapters.v1 JSON, then cassian preflight --adapter <adapter-output> to surface the IaC change context inside preflight findings. Both adapter output and preflight output are advisory and do not affect the gate verdict.
Recipe — What does my gate actually cover?
Demonstrates the advisory blast_radius.json artifact produced during cassian test's Collect phase. The recipe topology declares a scenario that touches only some of the available fault and wait classes, producing a meaningful coverage.summary block; blast radius does not affect the gate verdict.
Recipe — What coverage gaps do I have?
Demonstrates cassian preflight producing the advisory preflight.json artifact via declared-only static analysis (no deploy, no runtime). The recipe topology declares deliberate coverage gaps so the findings are meaningful; preflight does not affect the gate verdict.