Shipplane Lite
Turn the shell scripts you run by hand into workflows you can watch, re-run, and read the logs from. One binary, your own machine, no database to set up.
A workflow
Provisioning a fresh VPS, in your config file:
cmd is passed to the remote shell as-is, so a template literal keeps a multi-step command readable. Leading indentation is ignored by the shell.
Start shipplane-lite serve and that workflow is a graph in the UI. Run it, and each box turns green as it finishes, with the remote output streaming underneath. A step that exits non-zero stops the run and shows you which one — no scrolling back through terminal history to find where it broke.
One workflow per host, committed to git, and every box you own has a setup you can read and re-run.
Who it's for
One person running their own machines. Deploying a side project, provisioning a VPS, or any multistep process where "did that actually work?" is a question you have asked out loud.
Not built for teams operating a fleet — there are no users, roles, or audit logs.
What Lite does not do
- Schedule runs. You start a run from the UI, the CLI, or the HTTP API. For a schedule, call the CLI from cron or a systemd timer.
- Send notifications. Check results in the UI, or use the CLI's exit code to trigger your own alert.
- Track desired state. Lite runs the commands you give it. Re-running does whatever your script does, so write idempotent steps.
What's inside
Shipplane Lite (@shipplane/liteapp) wires together the Shipplane workflow packages into one deployable app:
@shipplane/coreruns the workflow graph, deciding task order fromdependsOnand handling retries, timeouts, cancellation, and logs.@shipplane/orchestratordispatches each task to an in-memory worker.@shipplane/visualizer-reactrenders the workflow graph and editor in the built-in UI.- An HTTP server exposes a JSON API and serves the UI.
- A CLI runs workflows and streams logs from the terminal using the same binary.
- SQLite (
better-sqlite3) stores workflow definitions, runs, and events. - A task registry provides two built-in task types and supports custom types through the config file.
- Lite sends no telemetry. Workflow definitions, runs, and logs remain in your local SQLite file.
Next steps
- Installation — install the binary and start the server.
- Configuration — HTTP host and port, config file, and workflow seeding.
- CLI — run workflows and follow logs from the terminal.
- HTTP API — the JSON API behind the UI.
- Registry — the two built-in task types and how to add your own.