Flux replays production bugs locally

This is the kind of debugging tool that can change incident response, because it turns a flaky production bug into a deterministic local repro instead of a log-chasing exercise.

• –`flux replay` is the wedge: it reuses recorded inputs without hitting live systems, so fixes can be validated without re-triggering side effects.
• –`flux resume` pushes the product beyond observability; it aims to continue the same execution after the bug is fixed, which is more ambitious than tracing alone.
• –The runtime is opinionated: JS/TS inside a controlled V8 isolate with external calls recorded to Postgres should give strong fidelity, but it narrows adoption to teams willing to run Flux’s stack.
• –Best fit is API-heavy backends where duplicate work is expensive, like payments, messaging, webhooks, and workflow orchestration.
• –Open source plus an Apache-2.0 license helps trust, because teams can inspect how determinism and replay guarantees are enforced.