From aidoc-flow
Creates a Product Requirements Document (PRD) from an upstream BRD, defining product features, personas, success metrics, and acceptance criteria at the C4 Container level. Use after BRD exists and before EARS.
How this skill is triggered — by the user, by Claude, or both
Slash command
/aidoc-flow:doc-prdThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Create a **Product Requirements Document (PRD)** — Layer 2 of the SDD flow.
Create a Product Requirements Document (PRD) — Layer 2 of the SDD flow. A PRD defines product features, user personas, success metrics, and acceptance criteria at the C4 Container level — what the product does, not how it is built.
Layer: 2 — Container level (product features and functional blocks). Upstream: BRD (Layer 1). Downstream: EARS → BDD → ADR → SPEC → TDD → IPLAN → Code.
Each PRD corresponds to one BRD iteration cycle (MVP → PROD → new MVP). New
scope gets a new PRD (PRD-02, PRD-03) rather than expanding an existing one;
link cycles with @depends: PRD-NN.
Use doc-prd when:
For end-to-end generation from a BRD, a prompt, or an IPLAN, use
../doc-prd-autopilot/SKILL.md.
PRD requires an upstream BRD. Before writing, read:
docs/01_BRD/BRD-NN_{slug}/), read all files as one logical
document.framework/layers/02_PRD/PRD-TEMPLATE.yamlframework/layers/02_PRD/README.mdframework/governance/ID_NAMING_STANDARDS.mdConfirm no ID collision: ls docs/02_PRD/ 2>/dev/null. Reference only BRD
elements that exist; never invent placeholders like BRD-XXX or TBD. The PRD
ID need not match the BRD ID (PRD-09 may implement BRD-16).
PRD is the Container level. Keep product capabilities, user journeys, and error handling; push everything else downstream.
| Content | Belongs in |
|---|---|
| Product features, personas, KPIs, journeys | PRD (this layer) |
WHEN-THE-SHALL-WITHIN formal requirements | EARS (Layer 3) |
Given-When-Then executable scenarios | BDD (Layer 4) |
| Architecture decisions | ADR (Layer 5) |
| Schemas, endpoints, infra config | SPEC (Layer 6) |
document_control comes first (product name, version, status, dates,
author/reviewer/approver, @brd: reference, EARS-Ready score, revision
history). Then:
See PRD-TEMPLATE.yaml for per-section content and embedded _guidance.
Hold role definitions and story summaries (As a [role], I want [capability] so that [benefit]) with product-level acceptance criteria only.
Include the note: detailed behaviors live in EARS; executable scenarios live
in BDD. Do not write WHEN-THE-SHALL or Given-When-Then here.
Must carry substantive content in at least 3 categories: product positioning, key messaging, feature descriptions, documentation, help text, error messages, success confirmations, onboarding, release notes. Placeholder-only content is a blocking error.
Elaborate BRD §7.2 topics with technical options and evaluation criteria. Layer separation: BRD = what & why · PRD = how to evaluate · ADR = the decision. Do not reference ADR numbers — ADRs do not exist yet.
PRD.{doc_id}.{section_id}.{hash} (e.g.
PRD.01.09.b3f2; hash = first 4 hex of SHA256 of
"{doc_id}:{section_id}:{title}:{description}" from PRD content, extend to 8
on collision). SS is the section the element lives in — no numeric
type-codes.@brd: tags (e.g.
@brd: BRD.01.07.a7f3). Downstream artifacts tag it: @prd: PRD.01.09.b3f2.FR-XXX, US-XXX, AC-XXX, F-XXX, and
the legacy 3-segment PRD.NN.xxxx.@threshold: PRD.NN.{category}.{key}
(categories: quota, risk, perf, timeout, rate).PRD-NN (two digits: PRD-01, PRD-99,
PRD-102).docs/02_PRD/PRD-NN_{slug}/ regardless of size. Monolithic:
PRD-NN_{slug}.md inside it; section-based (>25 KB): PRD-NN.S_{section}.md
framework/layers/02_PRD/PRD-00_index.TEMPLATE.md.@brd: tags resolving to existing BRD elements.docs/02_PRD/PRD-00_index.md and add this PRD to
the parent BRD's Downstream Artifacts in the same change.The framework ships no runtime code — this skill is the validator. Apply the
checklist against framework/layers/02_PRD/README.md and
framework/governance/ID_NAMING_STANDARDS.md.
@brd: reference and
EARS-Ready score.PRD.NN.SS.xxxx; SS equals the host section; no
removed patterns.@brd: tags resolve to existing BRD elements.@diagram: c4-l2 and @diagram: dfd-l2 present (use
../charts-flow/SKILL.md); sequence diagrams include alt/else.| Code | Meaning | Severity |
|---|---|---|
| XDOC-002 | Missing cumulative tag (@brd) | error |
| XDOC-006 | Tag format invalid | error |
| XDOC-008 | Broken internal link | error |
| XDOC-009 | Missing traceability section | error |
Quality gate (blocking): EARS-Ready score ≥ 90/100 before moving on. If issues are found, fix and re-check; if unfixable, log for manual review.
../doc-ears/SKILL.md — the EARS references this PRD (@prd: PRD.NN.SS.xxxx),
carries cumulative @brd/@prd tags, and formalizes PRD features into
WHEN-THE-SHALL-WITHIN requirements.
Before applying defaults, read the project adaptation profile
(.aidoc/profile.yaml). Honor only this skill's declared knobs:
section_toggles (include or omit template-declared optional sections)
and glossary (substitute preferred terms in generated prose). Ignore any
unknown or out-of-surface key; absent a profile, use framework defaults.
Authority: framework/governance/ADAPTATION.md.
framework/layers/02_PRD/PRD-TEMPLATE.yamlframework/layers/02_PRD/README.mdframework/layers/02_PRD/PRD-00_index.TEMPLATE.mdframework/governance/ID_NAMING_STANDARDS.md../doc-brd/SKILL.md../doc-prd-audit/SKILL.md · Fixes: ../doc-prd-fixer/SKILL.md../doc-prd-autopilot/SKILL.md| Purpose | Define product features, personas, and KPIs |
| Layer | 2 (Container) |
| Upstream tags | @brd (1 cumulative) |
| Key decision | What stays PRD-level vs pushes downstream |
| Must include | Document Control (first), §10 (≥3 categories), §14 ADR topics, 15 sections |
| Next | doc-ears |
npx claudepluginhub vladm3105/aidoc-flow-framework --plugin aidoc-flowAutomates full PRD lifecycle: initializes workflow, generates PRD via isolated Writer subagent, reviews via Reviewer subagent, iterates revisions up to 3 times until approval.
Generates structured PRDs with problem, context, solution, user stories, acceptance criteria, metrics, risks, and out-of-scope items. Iteratively gathers info via questions, reviews docs/issues/templates.
Generates Chinese PRD documents (full, brief, or one-page) via guided interviews, templates, validation, and AI-agent optimizations. Saves to docs/prd dirs.