Lightweight dev orchestrator for polyglot teams: typed
docker-composesuccessor with built-in dashboard, OpenTelemetry and production export.
LightShuttle (binary: lightshuttle) is a developer-time orchestrator written in Rust. You declare your service stack once in lightshuttle.yml (databases, queues, containers, Dockerfiles), and lightshuttle up boots the whole thing on your laptop with automatic service discovery, an integrated web dashboard, OpenTelemetry traces and logs, and a one-command export to docker-compose.yml, Kubernetes manifests or a Helm chart for production.
LightShuttle is sponsored by Nubster.
LightShuttle is published on crates.io and under active development. The public API is still pre-1.0 and may change between minor versions; see the SemVer policy.
- v0.1.0 Minimum viable orchestrator: typed manifest, topological startup, healthchecks, graceful shutdown, env-var service discovery, and the core CLI.
- v0.2.0 Dashboard and observability: a local web dashboard, live log
and event streaming, a bundled OpenTelemetry collector, Prometheus
metrics and the
restartcommand. - v0.3.0 Production export:
lightshuttle exporttodocker-compose.yml, Kubernetes manifests or a Helm chart. - v0.4.0 Secrets and env management:
.envsources,secrets check, fail-fast injection at boot, and a per-project Docker network.
See the roadmap for what comes next.
Prerequisites: Docker Desktop or a Docker Engine daemon running locally, plus a Rust toolchain (rustup is the recommended installer).
Install LightShuttle from crates.io:
cargo install lightshuttleTyping
lightshuttleis verbose. If your shell has nolshalready (check withcommand -v lsh), see the optional shell alias section of the tutorial.
Create a lightshuttle.yml file at the root of your project:
# lightshuttle.yml
project:
name: hello
resources:
db:
postgres:
version: "16"Boot the stack:
$ lightshuttle upThe orchestrator validates the manifest, pulls the image, starts Postgres, waits for the healthcheck to pass and supervises the container until you press Ctrl+C. Shutdown is coordinated and idempotent.
For a complete walkthrough see the getting started tutorial. Runnable manifests live in examples/.
docker-compose is the de facto local stack runner for polyglot teams, but it suffers from real ergonomic gaps that LightShuttle aims to close:
- No typing, a typo in a service name silently breaks the stack.
- No dashboard, you fall back to
docker logsanddocker psfor everything. - No service discovery, you manually wire
DATABASE_URLandREDIS_URLbetween containers. - No production export, the same YAML cannot be reused or transformed into Kubernetes resources.
- No custom lifecycle, running a migration before serving traffic requires shell scripts and
depends_on: service_healthyceremonies.
LightShuttle keeps the simplicity of a single YAML file but layers type validation, a live dashboard, automatic environment variable injection between services, and a first-class production export, with a clean Rust extension point for custom hooks and lifecycles on the roadmap.
To stay focused, the following are explicitly out of scope:
- Not a production runtime. LightShuttle is a developer-time orchestrator. The production export targets Kubernetes, Helm or plain
docker-compose, but LightShuttle itself does not run in production. - Not a Kubernetes replacement. We generate manifests; we do not run a control plane.
- Not a service mesh. Service discovery is provided by environment variables and per-project network hostnames; sidecar proxies, mTLS and traffic shaping are deliberately left to dedicated tools.
- Not a CI/CD pipeline. LightShuttle is orthogonal to existing pipeline tooling.
Everything lives in a single lightshuttle.yml, a typed declarative manifest readable by every developer regardless of their primary language. It declares a project, a set of resources (postgres, redis, container, dockerfile), their dependencies and ${...} interpolations, and optional dashboard, observability and export sections. Environment references resolve from .env files and the system environment, with fail-fast diagnostics at boot when a value is missing. The full schema is in the manifest specification; a JSON Schema for editor autocompletion ships at docs/spec/manifest-v0.schema.json. Custom lifecycle hooks through a Cargo-style xtask/ crate are on the roadmap, not yet implemented.
| Command | Purpose |
|---|---|
lightshuttle up |
Boot the stack and supervise it until interrupted. |
lightshuttle down |
Stop every managed container. |
lightshuttle ps |
List managed resources and their status. |
lightshuttle logs <resource> |
Stream a resource's logs. |
lightshuttle restart <resource> |
Restart one resource through the running control plane. |
lightshuttle validate |
Parse and validate the manifest without starting anything. |
lightshuttle manifest |
Print the resolved manifest. |
lightshuttle export <target> |
Generate docker-compose.yml, Kubernetes manifests or a Helm chart. |
lightshuttle secrets check |
List required and missing secrets without booting anything; honours --env-file. |
lightshuttle alias install |
Manage the optional lsh shell alias. |
The full documentation is published at https://nubster-opensources.github.io/lightshuttle/. The docs/ index maps every guide, specification and policy to its source.
- Tutorials: getting started, the dashboard, export.
- Specifications: manifest, control plane API, observability, export.
- Project policies: roadmap, SemVer, MSRV, release process, governance.
The workspace is split into published crates: lightshuttle (the CLI), lightshuttle-manifest, lightshuttle-spec, lightshuttle-runtime, lightshuttle-otel, lightshuttle-control, lightshuttle-secrets and lightshuttle-export.
LightShuttle is open source and pre-1.0; the public API may still change between minor versions. See CONTRIBUTING.md for the contribution model and the Contributor License Agreement, and SECURITY.md to report a vulnerability. Feedback and discussion on the direction are welcome.
LightShuttle is exercised by a cross-OS smoke job on every change. The support level differs because container support differs per CI runner.
| Tier | OS | Smoke coverage | CI gate |
|---|---|---|---|
| Tier 1 | Linux | Real container boot (up/down) plus offline CLI |
Blocking |
| Tier 2 | macOS, Windows | Offline CLI (validate, export) |
Advisory |
macOS runners ship no Docker daemon and Windows runners default to Windows containers, so booting the Linux containers a manifest describes is not testable there in CI. On those targets the smoke verifies the binary ports natively: manifest parsing, path handling and artifact emission. macOS and Windows graduate to blocking once they prove stable.
Licensed under either of:
- Apache License, Version 2.0 (LICENSE-APACHE)
- MIT License (LICENSE-MIT)
at your option.
Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in the work by you, as defined in the Apache-2.0 license, shall be dual-licensed as above, without any additional terms or conditions.
See CONTRIBUTING.md for details, including the Contributor License Agreement (CLA).
Copyright © Nubster.