Concepts overview
An overview of the core concepts that make up Pipemason.
Pipemason takes an intent — a feature description, a ticket, a design link, or a whole system to build — and walks it from analysis to a pushed branch with the rigor of a careful engineering team. A handful of concepts describe how that work is organized and how it advances. Each has its own page; this is the map.
Runs
A run is a single execution of the per-story pipeline: one unit of work becomes one branch and (optionally) one pull request. It is the smallest thing Pipemason executes end to end.
Programs
A program operates above runs. It takes a larger intent, decomposes it into a dependency graph of stories, and runs the per-story pipeline once per story in dependency order — useful for building or upgrading whole systems rather than a single feature.
Stories
Stories are the units of work a program is decomposed into. Each story is sized to flow through the pipeline cleanly and is executed by a run.
Domains
Domains partition work by technology surface — api, web, db, mobile, cloud, plus the game domains. Several phases run in parallel by domain so independent surfaces make progress at once.
Modes
Modes tell a program what kind of work it is doing — building from scratch (greenfield) or changing an existing system (brownfield) — which changes the planning steps a program performs before it starts iterating.
Phases
Within a single run, work moves through a sequence of phases from analyze to finalize. Each phase has a clear job and a gate it must pass before the next phase begins.
State machine
The state machine is the contract that governs how a run advances between phases, when it loops back to retry, and when it stops and escalates to you rather than guessing.
Where to go next
- New to Pipemason? Read Runs, then Phases, then the State machine in that order.
- Building a whole system or upgrade? Read Programs and Modes.
- Curious how parallel work is split? Read Domains.