Getting Started
The on-ramp for running Blueprint for the first time. Five minutes to a scaffolded portal you can build on.
1. What you need
-
Node.js >= 18. The CLI checks nothing else; it is dependency-free (hand-rolled arg parsing, no
node_modulesto install for the CLI itself). -
The Blueprint CLI. Install global or run on demand:
npm install -g @nino-chavez-labs/blueprint-cli # or, no install: npx @nino-chavez-labs/blueprint-cli <command>The binary is
blueprint. Current version is0.1.0. -
A target directory for the new initiative — an absolute path where the portal gets stamped (e.g.
~/projects/my-project). -
A browser sensor, only when you reach Stage 0 legibility against a running prototype. The default is
browse-tool; wire it later, not now. Skip it for scaffolding.
Verify the install:
blueprint --version # -> 0.1.0
blueprint --help
2. The mental model
Blueprint runs a 7-stage pipeline: Legibility (0) -> Research (1) -> Design Principles (2) -> Prototype (3) -> Fact-Check (4) -> Documents (5) -> Deploy (6), with Iterate (7) after stakeholders see it. Each stage gate is an executable reviewer — research-completeness-reviewer, doc-quality-auditor, prototype-smoke-runner, and the rest — that has to PASS before the next stage unblocks. You drive the producing stages with slash-command skills: /blueprint-research, /blueprint-prototype, /blueprint-validate, /blueprint-docs, /blueprint-deploy, /blueprint-triage.
Before any of that, three orthogonal choices set the shape of the work. Variant is what you’re doing: Greenfield (build a north-star prototype), Midstream (revise a live in-flight product), or Brownfield (audit + prescribe against a mature product). Tier is how far you’ve gone: Tier 0 (pre-portal scratch, <=1 week), Tier 1 (a portal exists — the default starting point), Tier 2 (portal plus live product surfaces). Pattern is the portal’s IA: Pattern A (platform-portal, multi-surface product with a 6-route contract and an audience switcher) or Pattern B (redesign-review-portal, current-state-vs-proposed comparison for a brownfield audit). Pick variant, tier, and pattern before you scaffold — the wrong pattern forces a rebuild.
3. Install and scaffold your first portal
blueprint init scaffolds a Pattern A portal at Tier 1. It delegates to the canonical stamper, which copies the template, substitutes a fixed token set, writes blueprint.yml, and runs a post-stamp grep to prove no source-project strings leaked in.
The CLI dispatcher passes all flags straight through to the stamper:
blueprint init \
--name=my-project \
--display-name="My Project" \
--repo-url=https://github.com/owner/my-project \
--tagline="One-line product tagline" \
--variant=greenfield \
--tier=1 \
--pattern=A \
--target=~/projects/my-project
Flag notes:
--nameis the project slug (used in package metadata, footer brand, file substitutions).--variantisgreenfield,midstream, orbrownfield.--tieris0,1, or2. Tier 1 is the default starting point for a serious initiative.--pattern=Ais the supported initial stamp today. Pattern B has no initial-stamp path yet — the stamper supports Pattern B only through--mode=restamp-chromeagainst an existing portal, not a fresh scaffold.--targetis the absolute path to the new initiative root.
4. What you just got
After the stamp runs, the output reports three things:
- copied (binary) — assets carried over verbatim (logos, images).
- stamped (text + substitutions applied) — every file where the source slug was replaced with your
--name/--display-name/--repo-url. - mechanical check: PASS — the post-stamp grep found no unexpected residual source strings. If it fails, it lists the residuals as a stamper bug, not your problem to patch.
The target directory now holds a Tier 1 Pattern A portal: the apps/portal/ Astro + React shell (canonical 6-route IA — overview, discover, try, build, operate, inspect, roadmap), the packages/ workspace it consumes (@blueprint/ui, @blueprint/design-tokens), and a stamped blueprint.yml declaring your variant, tier, and pattern. Content is placeholder by design — Tier 1 means the front door exists, not that it’s filled in.
Sanity-check the scaffold without building or deploying anything:
blueprint doctor --target=~/projects/my-project
doctor loads the config, every discovered reviewer, and runs portal conformance, then tells you what it did not check (full build, browser verification, deploy connectivity) so a green here is never mistaken for a deploy green.
Next: the full run
Scaffolding is Stage 0-to-1 plumbing. The Tutorial walks the whole pipeline end to end — /blueprint-research through each reviewer gate to /blueprint-deploy — on a real initiative. Start there once blueprint doctor reports PASS.