incubator

Front-of-funnel home for pre-product work in petry-projects: raw idea Discussions that aren't yet a product, the decision brief / PRD-lite an idea is judged on, and disposable POCs. When a POC proves out, the idea graduates to its own product repo — it does not grow up here.

This repo is the missing pre-product stage of the org ideation pipeline (petry-projects/.github-private → feature-ideation → idea-triage → idea:approved → initiative-planner). Approved products already own their PRDs in their own repos (_bmad-output/planning-artifacts/); this repo is where an idea lives before it earns a repo of its own.

Lifecycle

Stage	Gate	Where it lives
Raw idea capture	—	Discussions here (Ideas category) — for ideas not tied to an existing product repo
Enhancement + decision brief / PRD-lite (Mary, Analyst)	before idea:approved	ideas/<slug>/ in this repo
POC / spike (throwaway)	after idea:approved	pocs/<slug>/ in this repo
POC needing real CI / deploy / dependency isolation	—	a disposable standalone poc-<slug> repo
Graduation to a real product	after the POC proves out	a new product repo (full BMAD PRD by John, PM)

The hard rule: graduation is a boundary, not organic growth. A proven POC seeds a new product repo where the full PRD is written. Keeping the incubator small is what keeps dev-lead / pr-review pointed at real products.

Layout

ideas/
  <slug>/
    brief.md            # decision brief / PRD-lite — the go/no-go artifact
    market-research.md  # competitive / market context (Mary, Analyst)
    ost.md              # Mermaid opportunity-solution-tree
  _TEMPLATE/            # copy this folder to start a new idea
pocs/
  <slug>/               # throwaway spikes; archive or delete freely

How an idea moves through here

1. Capture — open a Discussion in the Ideas category (or one is filed for you). Ideas that clearly belong to an existing product go in that product's repo instead; cross-cutting or net-new ideas come here.
2. Brief — once an idea reaches "Ripe" in idea-triage, copy ideas/_TEMPLATE/ to ideas/<slug>/ and let Mary fill brief.md (problem, market context, JTBD, RICE/ICE, key assumptions + cheapest test, recommendation). The brief — not a bare idea — is what a human decides idea:approved on. Discovery precedes the PRD: the brief exists to inform the go/no-go, never to substitute for it.
3. Go/no-go — a human applies idea:approved (or declines). Declined ideas stay as a closed Discussion / archived folder — cheap to keep, easy to revisit.
4. POC — spin a spike under pocs/<slug>/ (default) or, only if it needs its own CI/deploy/deps, a disposable poc-<slug> repo. Prefer the cheapest testable artifact — a working spike often decides faster than more documents.
5. Graduate — if the POC proves out, create a new product repo from petry-projects/repo-template; the brief.md seeds John's full PRD there. Archive the incubator folder.

Conventions

• <slug> — short kebab-case, stable, matches the Discussion where practical.
• Ideas that die cost nothing — do not delete history; close/archive so the reasoning stays searchable for the next person (or agent).
• Nothing here is a product. No production dependencies, no user-facing deploys, no long-lived infrastructure. If it needs those, it's time to graduate.

---

Baseline scaffolding (workflows, CODEOWNERS, LICENSE, SECURITY) comes from petry-projects/repo-template; see BOOTSTRAP.md for the one-time per-repo setup and AGENTS.md for the org-standards pointer.