The lighter Kamal alternative

Levelrail vs Kamal: one binary, about 200 MB idle, Apache 2.0.

Levelrail control plane memory, idle, measured

~200 MB

Install
1 binaryplus one agent per extra node
License
Apache 2.0one edition, no license key
Ahead on
9 of 38capabilities, per Kamal's docs

The memory figure is one idle measurement of the Levelrail control plane (method). As of 2026-11-18.

01

What you gain over Kamal

9 capabilities where Levelrail is a plain yes and Kamal's own docs say partly or no. Full wording and sources are in the table.

Capabilities where Levelrail is ahead of Kamal
Capability Levelrail Kamal
No SSH: agents dial out Agents dial out over mTLS No SSH from your machine or CI to each server
Level-triggered reconciler with status reasons Desired state converged, status condition with reason per pass No None: it runs Docker commands per deploy
Metrics and log search built in Node-local metrics at 15s and full-text log search No Bring your own
One-click rollback with pinned images Prior images pinned so cleanup cannot remove them Partly Rollback to chosen version
Encrypted secrets for env vars Envelope-encrypted, masked in UI, injected at container create Partly kamal secrets helpers extract secrets from password managers
Request firewall, rate limits, bot protection Ordered rules, IP bans, bot handling, challenges and rate limits No Proxy options cover buffering, body size limit and timeouts
IAM, audit log and 2FA in the free edition Allow/Deny IAM policies, passkeys and TOTP, and an audit log Partly Per-server audit log of deploy steps, last 50 lines via kamal audit
Audit log of every action Every mutating request recorded Partly Per-server .kamal/app-audit.log of deploy steps, read with kamal audit
Deploy freezes and policies Cron-defined freeze windows per app with an audited override reason Partly kamal lock acquire can block deploys during a maintenance window
02

All 38 capabilities

Statuses for Kamal are what its own documentation states; "Not stated" means its docs do not say. For Levelrail, "Active" means the feature ships today, with beta and stable tags from the feature status page and the exact scope in each note.

Levelrail
YesActive
Kamal
YesPartlyNo

A rule marks the 9 rows where Levelrail is ahead.

Levelrail compared with Kamal, by capability
Capability Levelrail Kamal
Architecture 3
No SSH: agents dial out (Levelrail ahead) Yesbeta Agents dial out over mTLS; no inbound ports Source for Levelrail, No SSH: agents dial out No SSH from your machine or CI to each server Source for Kamal, No SSH: agents dial out
Level-triggered reconciler with status reasons (Levelrail ahead) Yes Desired state converged, status condition with reason per pass Source for Levelrail, Level-triggered reconciler with status reasons No None: it runs Docker commands per deploy Source for Kamal, Level-triggered reconciler with status reasons
Single control plane binary Yes One control plane binary, plus one agent per extra node Source for Levelrail, Single control plane binary Not applicable
Observability 3
Metrics and log search built in (Levelrail ahead) Yes Node-local metrics at 15s and full-text log search Source for Levelrail, Metrics and log search built in No Bring your own Source for Kamal, Metrics and log search built in
Crashloop diagnosis with last log lines Yes Crashloop detection surfaces the last log lines Source for Levelrail, Crashloop diagnosis with last log lines Not stated in their docs
Public status page Yesstable Opt-in public page with components, 90 day uptime bars, incidents, maintenance and RSS Source for Levelrail, Public status page Not stated in their docs
Deploys 10
One-click rollback with pinned images (Levelrail ahead) Yesstable Prior images pinned so cleanup cannot remove them. A 5xx spike or crashloop after cutover rolls back automatically (live-tested); the latency signal is beta Source for Levelrail, One-click rollback with pinned images Partly Rollback to chosen version; old containers pruned after 3 days Source for Kamal, One-click rollback with pinned images
Readiness-gated deploys (zero-downtime on one node) Yes Cutover only after the readiness probe passes. Live-tested on one node; multi-node is beta Source for Levelrail, Readiness-gated deploys (zero-downtime on one node) Yes Zero-downtime cutover with kamal-proxy Source for Kamal, Readiness-gated deploys (zero-downtime on one node)
Preview environments per pull request Yesstable Stable on GitHub, beta on GitLab and Bitbucket Source for Levelrail, Preview environments per pull request Not stated in their docs
Docker Compose apps Active Pre-built images from the dashboard; git deploys can build Source for Levelrail, Docker Compose apps Not stated in their docs
Git push deploys from GitHub, GitLab, Bitbucket Active GitHub, GitLab, Bitbucket and Gitea documented; GitLab and Bitbucket previews are beta (synthetic webhooks). A git push remote on your own server (no provider needed) is beta Source for Levelrail, Git push deploys from GitHub, GitLab, Bitbucket Not stated in their docs
Builds from source without a Dockerfile Active Railpack detects Node, Python, Go and more. Ruby and PHP are accepted but only plan generation is verified, no live build; Rust, Deno and Elixir are rejected Source for Levelrail, Builds from source without a Dockerfile Partly Dockerfile required by default; buildpacks only when configured explicitly Source for Kamal, Builds from source without a Dockerfile
Scheduled jobs Yes Per-app cron jobs with run history, logs, retries and alerts Source for Levelrail, Scheduled jobs Not stated in their docs
Deploy freezes and policies (Levelrail ahead) Yesstable Cron-defined freeze windows per app with an audited override reason; also gate image auto-update and scheduled deploys Source for Levelrail, Deploy freezes and policies Partly kamal lock acquire can block deploys during a maintenance window and a non-zero exit from pre-deploy or other hooks aborts the command (hooks can be skipped with --skip_hooks); no scheduled freeze windows or policy engine Source for Kamal, Deploy freezes and policies
CI deploys with no stored secret Active GitHub Actions only: a GitHub OIDC token is exchanged for a 15 minute token scoped to one app. Other CI uses an API token Source for Levelrail, CI deploys with no stored secret Not stated in their docs
Static site hosting Yes Static HTML from git served by the embedded ingress with TLS; drag-and-drop folder or zip deploy Source for Levelrail, Static site hosting Not stated in their docs
Data 3
Managed databases with scheduled backups Active Many engines, scheduled S3 backups. Postgres point-in-time restore is tested live; MySQL, MariaDB and MongoDB are beta Source for Levelrail, Managed databases with scheduled backups Partly Accessories run containers such as MySQL; no backups documented Source for Kamal, Managed databases with scheduled backups
Point-in-time database restore Active Postgres stable with an S3 compatible bucket; MySQL and MariaDB beta; MongoDB has no point-in-time restore Source for Levelrail, Point-in-time database restore Not stated in their docs
Encrypted secrets for env vars (Levelrail ahead) Yes Envelope-encrypted, masked in UI, injected at container create Source for Levelrail, Encrypted secrets for env vars Partly kamal secrets helpers extract secrets from password managers; custom TLS certs load from secrets; no encryption at rest documented Source for Kamal, Encrypted secrets for env vars
Networking 4
Private networking between apps or nodes Active Named networks and allow rules; cross-server mesh unverified on real hosts Source for Levelrail, Private networking between apps or nodes Partly Containers share a custom Docker network per host, no cross server networking Source for Kamal, Private networking between apps or nodes
Load balancing across replicas Active Apps with more than one replica are round-robin balanced by default, with failed-replica skip and retry. The configurable balancer (health checks, sticky sessions, weights, drain) is experimental, behind a flag, verified on one machine only. HTTP only Source for Levelrail, Load balancing across replicas No Proxy config covers hosts, health checks, timeouts and buffering for one server; no balancing across servers is described Source for Kamal, Load balancing across replicas
Request firewall, rate limits, bot protection (Levelrail ahead) Yes Ordered rules, IP bans, bot handling, challenges and rate limits Source for Levelrail, Request firewall, rate limits, bot protection No Proxy options cover buffering, body size limit and timeouts; no rate limits or bot protection described Source for Kamal, Request firewall, rate limits, bot protection
Automatic TLS certificates Active Self-signed by default; public ACME (HTTP-01) has one recorded run on a live VPS. Renewal near expiry and DNS-01 wildcards are not exercised live; CAs needing EAB are unsupported Source for Levelrail, Automatic TLS certificates Partly kamal-proxy Let's Encrypt HTTPS, only with one server and host set Source for Kamal, Automatic TLS certificates
Platform 11
Multi-node Activebeta Join, cordon and drain with real container relocation verified across two local Docker daemons; WireGuard mesh and cross-host transport still unverified. Resource-aware placement is beta Source for Levelrail, Multi-node Yes SSH to each server Source for Kamal, Multi-node
One-click service templates Yesbeta 592 catalog entries, 583 active; 346 proven on a real engine, 246 previews without that pass yet Source for Levelrail, One-click service templates Not stated in their docs
Autoscaling Active Beta: CPU, memory or request-rate scaling with cooldowns and reasons Source for Levelrail, Autoscaling Not stated in their docs
Sleep idle apps, wake on request Yesstable Opt in per app; stops after 5 minutes to 7 days idle and wakes on the next request. Verified live against a real Docker daemon and the real ingress Source for Levelrail, Sleep idle apps, wake on request Not stated in their docs
CLI and HTTP API Yes CLI and REST API documented; MCP server is beta Source for Levelrail, CLI and HTTP API Not applicable
Deploy and alert notifications Active Webhook and SMTP email have live e2e; 15 vendor kinds (Slack, Discord, PagerDuty and more) are beta, tested against local fakes Source for Levelrail, Deploy and alert notifications Partly Hooks run custom scripts; no built-in Slack, Discord or email Source for Kamal, Deploy and alert notifications
MCP / AI assistant integration Yesbeta MCP server (beta) with over 240 tools in toolsets an agent enables on demand (256 at last count, 21 in the default session); in-app chat is behind a flag Source for Levelrail, MCP / AI assistant integration Not stated in their docs
Runs on servers you own Yes Runs on Linux servers you provide Source for Levelrail, Runs on servers you own Yes Runs against servers you own over SSH Source for Kamal, Runs on servers you own
Open source license Yes Apache 2.0, one edition, no paid tier Source for Levelrail, Open source license Yes Released under the MIT License Source for Kamal, Open source license
Declarative apply of the whole platform Activebeta YAML export, plan and apply from CLI, dashboard, MCP or CI, plus a Terraform and OpenTofu provider. Experimental, behind a flag Source for Levelrail, Declarative apply of the whole platform Partly config/deploy.yml declares app, roles, proxy, accessories and servers and kamal setup/deploy applies it; commands run per deploy, no state reconciliation or drift detection documented Source for Kamal, Declarative apply of the whole platform
Functions and one-off jobs Active Node, Python or Go handlers that sleep when idle and wake on the first request; scheduled and background calls with retries Source for Levelrail, Functions and one-off jobs Partly A role can run a custom cmd such as bin/jobs for a job server; scheduled jobs are not described Source for Kamal, Functions and one-off jobs
Security 4
IAM, audit log and 2FA in the free edition (Levelrail ahead) Yesstable Allow/Deny IAM policies (stable, live e2e), passkeys and TOTP, and an audit log with a verifiable hash chain, CSV export and forwarding, all in the one edition Source for Levelrail, IAM, audit log and 2FA in the free edition Partly Per-server audit log (.kamal/app-audit.log) of deploy steps, last 50 lines via kamal audit; no users, roles or 2FA Source for Kamal, IAM, audit log and 2FA in the free edition
SAML or SCIM single sign-on Activebeta OIDC sign-in (17 named providers plus generic OpenID Connect) is on by default once configured. SAML and SCIM are beta, behind a toggle and off by default; SAML ran against Keycloak and SimpleSAMLphp in a local lab, not Okta or Entra Source for Levelrail, SAML or SCIM single sign-on Not applicable
Audit log of every action (Levelrail ahead) Yes Every mutating request recorded; verifiable hash chain, CSV or NDJSON export, retention, streaming to a webhook or S3 bucket Source for Levelrail, Audit log of every action Partly Per-server .kamal/app-audit.log of deploy steps, read with kamal audit; entries carry a user tag but the page does not say which person; hooks expose KAMAL_PERFORMER for custom reports Source for Kamal, Audit log of every action
Two-person deploy approvals Yesstable A different person with the deploy ability must approve; the requester cannot approve their own. Approval also works from Slack or Discord Source for Levelrail, Two-person deploy approvals Not stated in their docs

All ten projects are on the full matrix. Product names and logos belong to their owners and are used for identification only.

03

Move over from Kamal

No importer for Kamal. Point Levelrail at the same images and recreate configuration.

docker pull ghcr.io/glincker/levelrail:latest
04

Questions about Levelrail and Kamal

Is Levelrail a good alternative to Kamal?

Yes. Levelrail does what Kamal does with a lighter footprint and more built in. Desired state lives in a database, with a reconciler recording a status reason after every pass. Agents dial out, so CI uses a scoped API token instead of holding SSH keys. Adds a dashboard, deploy history, rollback and built-in observability, and is heavier by design. A read-only importer and a dry-run mode let you move without touching Kamal.

Can I move from Kamal to Levelrail?

No importer for Kamal. Point Levelrail at the same images and recreate configuration.

Is Levelrail ready to replace Kamal in production?

Run both side by side, move one domain at a time and keep Kamal as a fallback until you are happy. The feature status page lists the maturity of every area.

05

Keep exploring