What problem does it solve? Cluster behaviors of a realtime backplane cannot be proven by unit tests with a mocked hub. This Skill stands up a real multi-pod environment — several Nodefony applications sharing one Redis bus — to verify cross-pod fan-out, application isolation, bus injection resistance, and end-to-end latency and throughput. ## Core Features & Use Cases - Idempotent bench provisioning: setup.sh starts a Redis container, generates two applications linked to the local framework, installs a bench controller, and configures a shared backplane namespace; run.sh starts three pods (two sealed peers plus one unsealed control pod). - Attack matrix and measurement scripts: forge sealed attacker envelopes, inject raw Redis PUBLISH messages, and measure pure latency, burst throughput, publication cost, connection soak tiers, and memory peaks. - Database resilience benches: dedicated scripts test pod survival during database outages, readiness probe behavior on schema drift, and data persistence across graceful restarts. - Use Case: Verify that a message published on pod A1 reaches a WebSocket client on pod A2, that an unsealed injection is rejected by sealed pods but received by the unprotected control pod, and capture p50/p95/p99 latency across the bus. ## Quick Start Ask the AI to set up the multi-pod bench and run the cross-pod fan-out and injection tests using the nodefony-multipod-bench skill.
