§ reference · skill

blueprint

One-pass replacement for `design` + `plan` — decomposes a feature into vertical slices with developer micro-checkpoints and emits an implement-ready phased plan in a single run.

arguments [research artifact path]

§ 01 · purpose

Purpose

Mid-sized features where the architecture is not load-bearing enough to deserve a separate design pass, but a phased plan is still required. blueprint collapses decomposition and phasing into one skill, with checkpoints *between* slices so review happens mid-flight instead of after the whole plan is final.

§ 02 · when to use

When to use it

  • The change touches 6+ files but architecture is not the hard part.
  • You want iterative review between slices, not after the full plan lands.
  • You're starting from a research or explore artifact and want to go straight to an implement-ready plan.
  • Pick design + plan instead when architecture is genuinely load-bearing and deserves its own pass.

§ 03 · inputs

Inputs

Name Required Source
research artifact yes Path to .rpiv/artifacts/research/*.md or .rpiv/artifacts/solutions/*.md
task description no Free-text alongside the artifact path
research artifact
Open Questions seed the ambiguity queue; Developer Context Q/As are inherited decisions.

§ 04 · outputs

Outputs

Artifact Path Format
Implementation plan .rpiv/artifacts/plans/ markdown with - [ ] success-criteria checkboxes

§ 05 · key steps

Key steps

  1. Read research + key files into context Why: Same as design — the skill proceeds against real code, not against research's summary.
  2. Targeted depth research (parallel) Why: codebase-pattern-finder for code shape, optional web-search-researcher for novel work. Integration & precedent come from research itself — no rediscovery.
  3. Dimension sweep + holistic self-critique Why: Same six dimensions as design (data model · API · integration · scope · verification · performance) so a blueprint plan covers the same surface a design/plan pair would.
  4. Decompose into vertical slices, then generate slice-by-slice Why: Whole-feature decomposition first; per-slice code generation with developer micro-checkpoints between slices. Review interrupts the loop before it gets expensive to redirect.
  5. Finalize directly into .rpiv/artifacts/plans/ Why: Output is plan-shaped, not design-shaped — implement consumes it directly, no second pass needed.

§ 06 · related skills

upstream researchexplore
downstream implementvalidate