hashgraph-online/awesome-codex-plugins / plugins/alpoxdev/hypercore/skills/hyper-prd
hyper-prd
Verified[Hyper] Use this skill when the user wants a product requirements document for a web, iOS, Android, or desktop product - a new idea to size and write, an existing PRD to revise, or a PRD to review - and expects one `prd.md` file with goals, metrics, scoped stories, EARS requirements, acceptance scenarios, and a handoff block. It decides platform and size first, asks only the decisions the user alone can make within the size's question budget, marks unknowns inline, and stops at the document. Do not use it for user flows, wireframes, diagrams, HTML previews, technical design, task breakdowns, or implementation.
| File | Kind | Size |
|---|---|---|
| SKILL.ko.md | markdown | 23.3 KB |
| SKILL.md | skill md | 21.6 KB |
| assets/evals/hyper-prd-cases.jsonl | other | 12.9 KB |
| references/platforms/android.ko.md | markdown | 11.6 KB |
| references/platforms/android.md | markdown | 11.1 KB |
| references/platforms/desktop.ko.md | markdown | 8.9 KB |
| references/platforms/desktop.md | markdown | 8.6 KB |
| references/platforms/ios.ko.md | markdown | 9.6 KB |
| references/platforms/ios.md | markdown | 9.5 KB |
| references/platforms/web.ko.md | markdown | 10.3 KB |
| references/platforms/web.md | markdown | 10.1 KB |
| references/prd-template.ko.md | markdown | 25.0 KB |
| references/prd-template.md | markdown | 23.3 KB |
| rules/intake-and-questions.ko.md | markdown | 6.9 KB |
| rules/intake-and-questions.md | markdown | 6.7 KB |
| rules/requirements-writing.ko.md | markdown | 10.0 KB |
| rules/requirements-writing.md | markdown | 9.7 KB |
| rules/validation.ko.md | markdown | 8.7 KB |
| rules/validation.md | markdown | 8.2 KB |
rules/intake-and-questions.md
View on GitHub# Intake and Questions
Use this rule before the first question. It owns the platform and size judgment, the explore-first order, the question budget and format, and the defaults that are never worth asking about.
## 1. Decide the platform first
Target platforms are `web`, `ios`, `android`, `desktop`. More than one is normal. Decide from the request and the evidence, not from a template default.
| Signal | Platform |
|---|---|
| Browser, URL, responsive layout, SEO, session in a tab | `web` |
| App Store, TestFlight, an iPhone or iPad build | `ios` |
| Play Store, an APK or AAB, an Android tablet build | `android` |
| An installed application, tray icon, installer, auto-update, notarization, code signing | `desktop` |
Record the decision in the meta block with a one-line reason, for example `platforms: web, ios - the request names a browser product and an App Store release`. When the request is genuinely ambiguous between two platforms and the answer changes the scope materially, that is one question, not a guess buried in the document.
A platform decision is not a checklist. The platform block in `prd.md` records the decided platform and its open obligations; a platform checklist fills that block when one exists.
## 2. Decide the size
| Size | Use when | Effect on the document |
|---|---|---|
| `S` | One feature or one behavior change inside an existing product; one user-visible outcome; no new surface | One goal, one or two stories, a short requirement list, no P2/P3 sections unless they carry real scope |
| `M` | A new product surface or a multi-step flow; several stories that can ship in sequence | Full section set, P1/P2/P3 stories, several requirements, platform blocks for each decided platform |
| `L` | A new product, several platforms, or a change that alters the product model | Full section set plus explicit risks, dependencies, and the must-not-change behavior |
Record the size with a one-line reason: `size: S - one settings screen changes an existing flow`. When two sizes fit, take the smaller one. An oversized document is a defect: the failure mode this skill exists to avoid is a workflow that treats every idea like a product launch.
## 3. Explore before asking
Read, in this order, until the question would be answered:
1. the repository: existing product docs, README, configuration that names platforms or constraints
2. existing documents under `.hyper/prd/` for the same product, to avoid contradicting a decided PRD
3. the context the user pasted, including links, screenshots described in text, and prior conversation
Never ask what these sources answer. If a source answers part of a question, ask only the remainder. If two sources disagree, that disagreement is a finding for the document's open questions, not a question to the user unless it blocks scope.
## 4. Question budget and format
| Size | Questions |
|---|---|
| `S` | `S 1` |
| `M` | `M 3` |
| `L` | `L 5` |
Rules for every question:
- One question at a time. Never a numbered list of questions in one message.
- Two to five options, each concrete enough to choose without a follow-up.
- Exactly one option marked as recommended, with a one-sentence reason.
- Ask only what the user alone can decide: business priority, target audience, pricing intent, legal posture, launch scope, a tradeoff between two acceptable outcomes.
- Do not ask what the document can state as an assumption. If the answer is derivable from convention, take the convention and mark it.
- Silence, "알아서", "적당히", or an answer that does not pick an option means: take the recommended option, record it as `[가정: …]`, and continue. Do not re-ask. The marker sits in the sentence that uses the value: a value chosen by silence is never written as fact, even in a section far from the question.
- A question left unanswered and blocking (the document cannot be scoped without it) keeps `status: draft` and appears in the open questions section with the answerer named.
## 5. Defaults that are not worth asking
Take these unless the request contradicts them, and record the choice as an assumption only when it changes scope:
- language of the document body: the user's language, otherwise Korean
- slug: kebab-case from the product or feature name
- file location: `.hyper/prd/<slug>/prd.md`
- section order: the template order
- status: `draft` until the checklist passes with no blocking question
- priority scale: `P0` for a requirement without which the goal fails, `P1` for the first shippable scope, `P2` for a later requirement
- metric measurement time: the first measurement window the product can realistically observe, stated explicitly rather than left blank
- platform checklist content: recorded as an open obligation, never invented
## 6. Report the intake
Before writing sections, state in one short block: platforms with reason, size with reason, how many questions the budget allows, and which sources were read. This block goes into the meta area of `prd.md`, not into a separate file.
## Sources
> Checked 2026-11-07.
| Claim | Source |
|---|---|
| A bounded question count instead of an open interview | [snarktank/ai-dev-tasks create-prd](https://raw.githubusercontent.com/snarktank/ai-dev-tasks/main/create-prd.md) - "Limit questions to 3-5 critical gaps in understanding." |
| One question at a time and a session-wide cap | [spec-kit clarify](https://raw.githubusercontent.com/github/spec-kit/main/templates/commands/clarify.md) - "Maximum of 5 total questions across the whole session." / "Present EXACTLY ONE question at a time." |
| A list of reasonable defaults that should not be asked about | [spec-kit specify](https://raw.githubusercontent.com/github/spec-kit/main/templates/commands/specify.md) - "Examples of reasonable defaults (don't ask about these)" |
| Discover facts instead of asking for them | [OpenAI Codex plan template](https://raw.githubusercontent.com/openai/codex/main/codex-rs/collaboration-mode-templates/templates/plan.md) - "Eliminate unknowns in the prompt by discovering facts, not by asking the user." |
| Ask only what changes the architecture or the decision | [Citadel prd](https://github.com/SethGammon/Citadel) - "Ask up to 3 questions — only those that would change the architecture." |
| Platform-specific obligations belong in the PRD from the start | [strvcom flutter-template prd](https://github.com/strvcom/flutter-template) |
| A document proportional to the problem | [Martin Fowler, SDD tool comparison](https://martinfowler.com/articles/exploring-gen-ai/sdd-3-tools.html) - "it quickly became clear that the workflow was like using a sledgehammer to crack a nut" |
### Evidence grade
`LOCAL` - the platform table, the size table, and the defaults list are this package's design. The `S 1, M 3, L 5` budget is a chosen value assembled from the cited caps, not a measured optimum.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| brainstormingobra/superpowersYou MUST use this before any creative work - creating features, building components, adding functionality, modifying behavior, or planning anything new, in software or out of it (a talk, a business, a renovation). | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| brand-guidelinesanthropics/skillsApplies Anthropic's official brand colors and typography to any sort of artifact that may benefit from having Anthropic's look-and-feel. Use it when brand colors or style guidelines, visual formatting, or company design standards apply. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| frontend-designanthropics/skillsGuidance for distinctive, intentional visual design when building new UI or reshaping an existing one. Helps with aesthetic direction, typography, and making choices that don't read as templated defaults. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| mcp-builderanthropics/skillsGuide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node/TypeScript (MCP SDK). | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |