Who Levelrail is for (and who it isn't)

[thegdsks]

Positioning honestly, including the "don't use this" cases, since that's more useful than pretending Levelrail fits everyone.

Levelrail is a reasonable fit if

• You're running somewhere between 3 and 50 services across 1 to 10 Linux boxes.
• You want push-to-deploy with TLS, logs, metrics, and rollback, without SSH-and-shell-scripts as the deploy mechanism.
• You want metrics and log search included by default, not a separate Grafana/Loki install you're expected to wire up yourself.
• You're comfortable running pre-1.0 software where the API and app spec can still change.

Levelrail is probably the wrong tool if

• You need Kubernetes's actual scale or ecosystem (operators, CRDs, a scheduler doing real bin-packing). This project explicitly won't grow into that: no CRDs, no custom orchestration standard, no autoscaling in v1. If you need those, use Kubernetes.
• You need Windows or non-Linux node support. Not planned.
• You need a managed cloud offering today. There isn't one; self-hosted only, and a managed offering is explicitly deferred until the self-hosted version has real users.
• You need a huge one-click template catalog. There's a curated, ~15-entry catalog today (browsable from the app creation wizard), not Coolify's 280; ten to twenty curated, actually-tested templates was always the target, not raw count.
• You need production stability today. There's no stable release yet.

The actual wedge

Coolify, Dokploy, and CapRover drive remote servers over SSH, shelling docker or docker compose commands and parsing text output, which is both a common source of flakiness and forces polling loops. Levelrail's agent talks to the Docker Engine API directly, no CLI shelling anywhere. That plus observability being a core primitive instead of a bolt-on is the actual bet this project is making. Full technical comparison: docs/comparison.md.