Skip to content

Guides overview

Practical, end-to-end walkthroughs for the workflows you'll reach for most: shipping a single feature, building a whole system, upgrading an existing one safely, and keeping spend predictable along the way.

Ship a single feature

The fastest way to feel how Pipemason works is to point it at one small change inside a git repo and watch it go end-to-end — branch, tests, implementation, gates, and an optional pull request.

  • Your first run — create and execute a single-story run with pipemason start, then watch it live on the dashboard.
  • Handling escalations — what to do when the runner pauses and asks for a human decision instead of guessing.

Build and evolve whole systems

A program runs above a single feature. You describe an intent, a plan is produced as a dependency graph of stories, and each story runs the full per-story pipeline in order.

  • Multi-story program — generate an entire system from a single greenfield intent.
  • Brownfield upgrade — change an existing system across multiple stories with migration discipline.

Operate and integrate

  • Cost control — set per-run, cumulative, and per-day ceilings so a run can never spend more than you expect.
  • Using the dashboard — monitor runs and programs as they happen from the web app.
  • GitHub integration — connect a repository once for automated branch and pull-request management.

Tip

New to Pipemason? Start with Your first run, then come back here once you want to scale up to programs.

Before you start

Every guide assumes you've installed the CLI and paired a runner. If you haven't yet, do that first:

npm install -g @pipemason/cli
pipemason login

All commands below are run from inside a git repository on your own machine. Your source code never leaves it — only run metadata, logs, and any artifacts you choose to share are streamed to the dashboard.