Pipelines

Pipelines let you chain multiple commands into a named, reusable workflow. Each step references a command from any tool (Supabase, GitHub, Vercel, or Pulumi) with its own args and flags.

Creating a Pipeline

Use the Pipeline Builder screen in interactive mode to create a pipeline:

1. Name the pipeline and add an optional description
2. Add steps — each step picks a command from the board
3. Configure args, flags, and whether to continue on error
4. Save to project or global scope

Pipeline Steps

Each step has:

Steps run sequentially. If a step fails and continueOnError is false, remaining steps are skipped.

Storage

Pipelines can be stored at two levels:

• Project — saved in .polter/config.json at the repository root. Shared via version control.
• Global — saved in OS-level app config. Available across all repositories.

Project pipelines take precedence when both levels define a pipeline with the same ID.

Headless Execution

Run a pipeline without the interactive UI:

polter pipeline run <name>

This resolves the pipeline by name (project scope first, then global) and executes all steps sequentially, printing progress to stdout.

Example

A "deploy" pipeline that pulls migrations, deploys functions, and pushes to Vercel:

If step 1 fails, steps 2 and 3 are skipped (unless continueOnError is set).