LayerRail is an open source cloud control plane for developers and teams. Manage
virtual machines, networking, PostgreSQL, Kubernetes, GitHub runners, and AI
inference through a project-centered dashboard and API, with access controls,
resource state, and usage close at hand.
This repository contains the API, dashboard, and provisioning workers behind
LayerRail. Use the hosted platform to build, or explore the code and run the
control plane yourself.
See LayerRail in motion
Watch the launch film and open the interactive preview →
An introduction to the API, dashboard, and project workflow. The preview uses an
illustrative workspace; it contains no customer data.
From an API request to a running project
| Build with | What you can do |
|---|
| Compute | Create Linux virtual machines and inspect their state, sizing, networking, and backups. |
| Networking | Organize private subnets, control traffic with firewalls, and route requests through load balancers. |
| PostgreSQL | Manage databases, backup and restore workflows, configuration, and monitoring. |
| Kubernetes | Manage clusters, node pools, kubeconfigs, resizing, and upgrades. |
| GitHub Actions | Connect repositories to managed runners from the same project. |
| AI inference | Explore models in the playground and call supported chat, Responses, and embedding APIs. |
| Project operations | Manage team access, personal access tokens, audit logs, usage, credits, and invoices. |
Service availability depends on the deployment's enabled features, configured
providers, and model catalog. The hosted dashboard shows the options available
to your project.
Automate through the API
Create a personal access token in your project's dashboard, copy the project
ID, and list its virtual machines:
export LR_TOKEN="pat-YOUR_PERSONAL_ACCESS_TOKEN"
export PROJECT_ID="YOUR_PROJECT_ID"
curl --fail-with-body \
"https://api.console.layerrail.com/project/${PROJECT_ID}/vm" \
-H "Authorization: Bearer ${LR_TOKEN}"
Requests respect your account and token permissions. The
OpenAPI specification documents infrastructure endpoints,
request bodies, and pagination. For terminal workflows, see the lr CLI.
Bring AI into the same workflow
Create a separate inference API key in your project, then make a supported
Responses request:
export LAYERRAIL_AI_KEY="YOUR_INFERENCE_API_KEY"
curl --fail-with-body https://api.console.layerrail.com/v1/responses \
-H "Authorization: Bearer ${LAYERRAIL_AI_KEY}" \
-H "Content-Type: application/json" \
-d '{
"model": "gpt-6.1-sol",
"input": "Explain what a cloud control plane does in one paragraph.",
"reasoning": {"effort": "low"},
"max_output_tokens": 1024,
"store": false
}'
AI requests require an available model and verified project billing. Usage is
paid; model prices and supported request formats are shown in the catalog. See
the GPT-6 API notes and usage billing guide.
Try the dashboard locally
With Git, Docker Compose, and a POSIX shell installed:
git clone https://github.com/layerrail/layerrail.git
cd layerrail
./demo/generate_env
docker compose -f demo/docker-compose.yml up --build
Open localhost:3000. The demo starts a PostgreSQL
database, applies migrations, and runs the web service and background workers.
On Windows, run these commands from WSL or another POSIX shell.
The demo provides a local control plane. Configure provider credentials and
service prerequisites before provisioning cloud resources. Follow the
getting started guide for account setup, API access,
and deployment configuration.
How it fits together
flowchart LR
Developer[Developers and teams] --> Dashboard[Dashboard]
Developer --> API[API and CLI]
Dashboard --> Control[Control plane]
API --> Control
Control <--> State[(PostgreSQL state)]
State <--> Workers[Provisioning and monitoring workers]
Workers --> Providers[Configured providers]
Providers --> Resources[Project infrastructure]
Control --> Inference[AI provider APIs]
The Ruby control plane uses Roda, Sequel, and Rodauth. The dashboard
uses server-rendered views with Tailwind CSS and a React-based AI playground.
Background workers provision and reconcile infrastructure; PostgreSQL stores
control-plane state.
Build with us
Read CONTRIBUTING.md before opening a pull request, and
DEVELOPERS.md for the development environment and test suite.
We welcome focused improvements to provider integrations, reliability,
documentation, and the developer experience.
Open source foundation
LayerRail is built on the Ubicloud open
source infrastructure foundation. This product fork retains the upstream
AGPL-3.0 license and attribution. Preserve the original notices
and meet the license's source-sharing requirements when modifying or operating
the software over a network.
