Not every integration ships with a REST API. Sometimes the truth lives in a bash script IT wrote in 2014, a vendor CLI, or a one-off Python utility that “just works.” Rip-and-replace is expensive. Bridge-and-automate is faster.
Nodlyn’s Command Line connector lets workflows invoke host programs and capture stdout, stderr, and exit codes — so your canvas orchestrates what already runs on the machine, with the same validation, approval, and audit trail as any other node.
Why teams bridge to the shell
- Legacy tooling — ffmpeg, rsync, custom exporters, internal CLIs nobody wants to rewrite.
- Edge & on-prem — commands run where the files and devices are, not in a distant cloud region.
- Gradual modernization — wrap a script today; replace it with native nodes tomorrow without losing the surrounding workflow.
The connector is marked Dangerous on purpose. Shell access is power. Nodlyn pairs it with approval gates and lifecycle controls so production runs are deliberate, not accidental.
Governance without slowing innovators
Engineers prototype on the canvas with Dry Run. Supervisors approve graphs that arm dangerous nodes. Operators see every command, exit code, and duration in run history — not a black box cron entry on a server nobody monitors.
That is the difference between “we run scripts” and “we run governed automation.”
Real patterns we see
Health checks — ping or curl internal services on a schedule; branch on exit code; page on-call if something fails.
Media & files — trigger transcoding or archival when a file lands in a watched folder (often combined with file triggers and HTTP notifications).
DevOps glue — run deployment helpers, rotate logs, or invoke git-based tooling as part of a larger approval workflow.
Arguments can pull from upstream data and shared config — so the same graph works across environments without hard-coding paths. See Global Variables for how teams centralise those values.
Local-first fits shell automation
Cloud automation platforms often cannot run arbitrary commands on your LAN. Nodlyn can — because execution stays on your hardware or on a runtime agent you control.
Technical reference (properties, outputs, timeout, shell mode): Command Line in the Knowledge Base.
Keep the tools you trust. Add the visibility and control you were missing.