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 |
references/prd-template.md
View on GitHub# PRD Template
Use this reference when creating or restructuring `prd.md`. It owns the section order, the skeleton, and one worked example.
## Contents
- Section order
- Skeleton
- Worked example
- Release gate table
- Data collection matrix
- Rules
## Section order
1. Meta: status, platforms, size, questions asked of the budget, sources read, updated
2. One-line definition
3. Problem and evidence
4. Target users and explicit exclusions
5. Goals and metrics
6. Scope: P1/P2/P3 stories, non-goals, must-not-change behavior
7. Functional requirements
8. Acceptance scenarios
9. Platform requirements: one block per decided platform, a release gate table in each block, and the shared data collection matrix when both `ios` and `android` are targets
10. Assumptions
11. Open questions
12. Risks and dependencies
13. Change history
14. Handoff block
Delete a section that does not apply. Never keep a heading without content, and never write `N/A`.
## Skeleton
```markdown
# <Product or feature name> PRD
## Meta
- Status: draft | ready | revised
- Platforms: web | ios | android | desktop (one or more) - <one-line reason for the decision>
- Size: S | M | L - <one-line reason for the decision>
- Questions: <asked>/<budget>
- Sources read: <repository paths, documents, and pasted context read before the first question>
- Updated: YYYY-MM-DD
## Definition
<One sentence. What this is, for whom, and what changes.>
## Problem and evidence
- One concrete case: <who, when, what happened, with real values>
- Current alternative: <what people do today instead>
- Evidence grade: observed | reported | assumed
## Target users and exclusions
- Users: <the people this serves, specifically>
- Excluded: <who this deliberately does not serve>
## Goals and metrics
| Goal | Metric | Target | Measurement means | Measurement time |
|---|---|---|---|---|
| <goal> | <primary metric> | <number + unit> | <where the number comes from> | <window> |
| <goal> | <secondary metric> | <number + unit> | <source> | <window> |
| <goal> | <counter-metric> | <floor + unit> | <source> | <window> |
## Scope
### P1 - <story name>
- Priority reason: <why now, what breaks if it slips>
- Independent test: <how this story alone is proven>
- Requirements: FR-001, FR-002
### P2 - <story name>
- Priority reason: <...>
- Independent test: <...>
- Requirements: FR-003
### P3 - <story name>
- Priority reason: <...>
- Independent test: <...>
- Requirements: FR-004
### Non-goals
- <A thing this deliberately does not do, and what it would have added>
- <A second thing, cutting a different part of the scope>
### Must not change
- FR-010 WHEN <existing condition>, the system SHALL CONTINUE TO <existing behavior>.
## Functional requirements
| Id | Priority | Requirement | Verification |
|---|---|---|---|
| FR-001 | P0 | <EARS sentence> | <what a reviewer does and sees> |
| FR-002 | P1 | <EARS sentence> | <...> |
## Acceptance scenarios
### <Story name>
```text
Scenario: <happy path name> (FR-001, FR-002)
Given <concrete starting state>
When <one action>
Then <observable outcome>
Scenario: <failure or edge name> (FR-002)
Given <concrete starting state>
When <one action>
Then <observable outcome>
```
## Platform requirements
### web
<One block per decided platform, in the order the platforms were decided. Name the checklist file the block was filled from. A platform that was not decided gets no block and its checklist is never read.>
#### Release gate
| ID | Grade | Item | Conclusion | Evidence | Re-check |
|---|---|---|---|---|---|
| W5 | `[M-조건]` | <the item sentence, short form> | applies, or not applicable with the condition, or `[확인 필요: <question>]` | <document, screen, or measurement that settles it> | <the date the dated value is checked again, or `none`> |
| W6 | `[M-조건]` | <...> | <...> | <...> | <...> |
| W1 | `[G]` | <only when relevant to this product> | <...> | <...> | <...> |
### Data collection matrix (ios and android)
Write this once, inside the platform requirements section and outside any single platform block, only when the decided platforms include both `ios` and `android`.
| Data type | Collected | Shared with third parties | Linked to the user's identity | Used for tracking | Purpose | Optional | Ephemeral |
|---|---|---|---|---|---|---|---|
| <e.g. email address> | yes | no | yes | no | account sign-in | no | no |
| <e.g. coarse location> | no | - | - | - | not collected; named so both stores agree | - | - |
Apple privacy label view: <which rows of the table map to which App Privacy label data types, and what the label says for each>.
Play Data safety view: <which rows map to which Data safety data types, and what the declaration says for each, including "no data collected" when that is the answer>.
- The matrix is written once. The App Privacy label (`I4`) and the Play Data safety declaration (`A2`, `A3`) both read from it; neither restates the data types from memory.
- A data type one platform handles and the other does not is a product decision: record the difference and its reason next to the matrix.
## Assumptions
- A-001: <statement>
- Evidence grade: repo fact | convention default | guess
- Basis: <path, quote, or convention>
- If wrong: <what changes>
## Open questions
- Q-001: <question>
- Answerer: <role>
- Blocking: yes | no
- History: Q: <question> → A: <answer or unanswered>
## Risks and dependencies
- Risk: <what could fail, and the early signal>
- Dependency: <what this needs from outside this document>
## Change history
| Date | Mode | What changed |
|---|---|---|
| YYYY-MM-DD | create | Initial document |
## Handoff
- Platforms: <decided platforms>
- Size: <S | M | L>
- P1 scope: <the stories and requirements that ship first>
- Remaining blocking questions: <Q ids, or "none">
- Inputs for the next stage: <what the next stage reads from this document>
```
## Worked example
```markdown
# Focus Timer PRD
## Meta
- Status: draft
- Platforms: ios, android - the support thread reports a phone timer, and the request names both stores
- Size: S - one screen and one behavior change on a new single-purpose app
- Questions: 1/1
- Sources read: the support thread quoted in Problem and evidence; no repository documents (new product)
- Updated: 2026-11-07
## Definition
A single-purpose focus timer that starts a fixed session and blocks mid-session edits, for people who lose the thread when a timer is too configurable.
## Problem and evidence
- One concrete case: a support thread on 2026-10-28 reports a user restarting a 25-minute session four times because the settings screen offered six duration presets mid-session.
- Current alternative: a general timer app plus a paper note of the intended duration.
- Evidence grade: reported
## Target users and exclusions
- Users: individual knowledge workers who run several short focus blocks a day
- Excluded: teams needing shared session dashboards; people who need interval or pomodoro chaining
## Goals and metrics
| Goal | Metric | Target | Measurement means | Measurement time |
|---|---|---|---|---|
| A session runs to completion | completed-session rate | 70% or more of started sessions | session start and end events | first 14 days after release |
| Sessions start fast | median time from app open to session start | 5 seconds or less | start-timestamp event pairs | first 14 days after release |
| The block does not cost the rest of the app | crash-free session rate | 99.5% or more (counter-metric floor) | crash report per session | first 30 days after release |
## Scope
### P1 - Start and finish one session
- Priority reason: without a run-to-completion path there is no product; nothing else is measurable.
- Independent test: start a 25-minute session, background the app, return after 25 minutes, see the completion state.
- Requirements: FR-001, FR-002, FR-010
### P2 - Resume after an interruption
- Priority reason: a killed process is the most common way a session dies; shipping resume doubles the useful sessions.
- Independent test: force-quit mid-session, relaunch, see the remaining time restored.
- Requirements: FR-003
### Non-goals
- Preset library and custom duration authoring: would add a settings surface and a second mental model for the same 25 minutes.
- Cross-device session sync: would add an account and a server dependency that the single-device case does not need.
- Paid tier: a subscription or one-off purchase would add store billing, a price surface, and an entitlement state the free timer does not need.
### Must not change
- FR-010 WHEN the app returns from the background during a session, the system SHALL CONTINUE TO show the remaining time without restarting the session.
## Functional requirements
| Id | Priority | Requirement | Verification |
|---|---|---|---|
| FR-001 | P0 | WHEN the user starts a session, the timer SHALL begin counting down from the fixed duration and show the remaining time. | Start a session; the remaining time decreases each second. |
| FR-002 | P0 | IF the session reaches zero, THEN the timer SHALL show the completion state and stop counting. | Run a one-minute test session to zero; the completion state appears and the count stops. |
| FR-003 | P1 | WHEN the process is killed mid-session and relaunched within 24 hours, the timer SHALL restore the remaining time. [가정: users return within a day] | Kill the process at 10 minutes remaining, relaunch, see 10 minutes. |
| FR-004 | P1 | WHERE the device is in silent mode, the completion state SHALL appear without an audible alert yet still show a visible banner. [확인 필요: is a haptic required by the design review?] | Enable silent mode, run to zero, see the banner and no sound. |
| FR-010 | P1 | WHEN the app returns from the background during a session, the system SHALL CONTINUE TO show the remaining time without restarting the session. | Background the app mid-session for two minutes, return, and see the countdown continue from the reduced remaining time. |
## Acceptance scenarios
### Start and finish one session
```text
Scenario: session runs to completion (FR-001, FR-002)
Given a fresh install with no session in progress
When the user taps Start and waits 25 minutes
Then the completion state is shown and no countdown remains
Scenario: session survives a background return (FR-001, FR-010)
Given a session started 10 minutes ago
When the user backgrounds the app for two minutes and returns
Then the remaining time shows about 13 minutes and the session is still running
```
### Resume after an interruption
```text
Scenario: a force-quit restores the remaining time (FR-003)
Given a session with 10 minutes remaining
When the user force-quits the app and relaunches within 24 hours
Then the timer shows the restored remaining time and keeps counting
Scenario: a relaunch outside the window starts a new session (FR-003)
Given a session that ended 25 hours ago
When the user relaunches the app
Then no session is restored and the start state is shown
```
## Platform requirements
### ios
Filled from `references/platforms/ios.md`.
#### Release gate
| ID | Grade | Item | Conclusion | Evidence | Re-check |
|---|---|---|---|---|---|
| I1 | `[M-조건]` | In-app account deletion under 5.1.1(v) | not applicable - the app creates no account and has no sign-in, so 5.1.1(v) has no subject | product scope above: no account, no sync | 2027-02-07 |
| I2 | `[M-조건]` | In-app purchase under 3.1.1 | not applicable - no paid digital goods or services | non-goals: no paid tier | 2027-02-07 |
| I3 | `[M-조건]` | Subscription terms under 3.1.2 | not applicable - no subscription | non-goals: no paid tier | 2027-02-07 |
| I4 | `[M]` | App Privacy label for every data type the release handles | applies - session events and crash reports, neither linked to identity nor used for tracking; the label reads from the data collection matrix below | data collection matrix | 2027-02-07 |
| I5 | `[M]` | Privacy manifest content and its owner | applies - manifest lists the same data types as the label, no tracking domains, owner is the release owner | data collection matrix | 2027-02-07 |
| I6 | `[M-조건]` | Review demo account under 2.1 | not applicable - no sign-in, so the reviewer needs no credentials | product scope: no account | 2027-02-07 |
| I7 | `[M]` | Minimum functionality under 4.2 and 4.2.2 | applies - the timer, its session state, and its completion state are the app's own function, not a repackaged site or a link collection | scope P1 | 2027-02-07 |
| I8 | `[G]` | Notification permission request in context | applies - the completion alert asks on first session start, with one sentence naming the timer | scope P1 | 2027-02-07 |
### android
Filled from `references/platforms/android.md`.
#### Release gate
| ID | Grade | Item | Conclusion | Evidence | Re-check |
|---|---|---|---|---|---|
| A1 | `[M]` | Target API level of the release | applies - the release targets API level 36, the floor in force from 2026-08-31 | build config at release time | 2027-02-07 |
| A2 | `[M]` | Play Data safety declaration | applies - the declaration is written from the data collection matrix below | data collection matrix | 2027-02-07 |
| A3 | `[M]` | Per-type collected, shared, purpose, optional, ephemeral | applies - session events and crash reports, not shared, not optional, not ephemeral | data collection matrix | 2027-02-07 |
| A4 | `[M]` | Permission list limited to promoted current features | applies - the notification permission serves the completion alert promoted in the listing | scope P1 | 2027-02-07 |
| A5 | `[M]` | Permission-to-feature mapping table | applies - one row: notification permission, completion alert, listing screenshot 2, asked on first session start | store listing plan | 2027-02-07 |
| A6 | `[M-조건]` | Sensitive or restricted permission declaration | not applicable - the release requests no sensitive or restricted permission | permission list above | 2027-02-07 |
| A7 | `[M-조건]` | Play billing for digital goods | not applicable - no digital goods are sold | non-goals: no paid tier | 2027-02-07 |
| A8 | `[M-조건]` | Play Billing library version | not applicable - the app does not use Play Billing | non-goals: no paid tier | 2027-02-07 |
| A9 | `[M]` | Published artifact format | applies - an Android App Bundle is uploaded, with no delivery split relied on | build config at release time | 2027-02-07 |
| A10 | `[G]` | Stability floor read from Android vitals | applies - the counter-metric reads user-perceived crash rate below 1.09% and ANR rate below 0.47%, per app version, in Play Console | goal 3 in the metrics table | 2027-02-07 |
| A11 | `[G]` | Interactive touch targets at 48 dp or more | applies - the start and stop controls are measured at 56 dp on one compact and one large screen | layout inspection record | 2027-02-07 |
### Data collection matrix (ios and android)
| Data type | Collected | Shared with third parties | Linked to the user's identity | Used for tracking | Purpose | Optional | Ephemeral |
|---|---|---|---|---|---|---|---|
| Session start and end events | yes | no | no | no | measuring the completed-session rate | no | no |
| Crash reports | yes | no | no | no | stability | no | no |
| Account identifiers | no | - | - | - | no account exists | - | - |
| Precise location | no | - | - | - | not requested | - | - |
Apple privacy label view: session events and crash reports are declared as usage and diagnostics, not linked to identity, not used for tracking; the label names no other type.
Play Data safety view: the same two types are declared as collected, not shared, not optional, not ephemeral; the declaration names no other type and does not claim "no data collected".
## Assumptions
- A-001: users accept a fixed 25-minute duration with no in-session settings
- Evidence grade: convention default
- Basis: the cited support thread is about too many presets
- If wrong: P1 gains a duration choice and the settings non-goal is removed
## Open questions
- Q-001: does the design review require a haptic on completion?
- Answerer: design
- Blocking: no
- History: Q: is a haptic required? → A: unanswered
## Risks and dependencies
- Risk: background execution limits stop the countdown; the early signal is a session that jumps on return.
- Dependency: the store listing needs the final duration copy before submission.
## Change history
| Date | Mode | What changed |
|---|---|---|
| 2026-11-07 | create | Initial document |
## Handoff
- Platforms: ios, android
- Size: S
- P1 scope: FR-001, FR-002, FR-010 with the start-and-finish story
- Remaining blocking questions: none
- Inputs for the next stage: the fixed duration, the completion state, the resume window, the iOS and Android release gate tables, and the shared data collection matrix
```
## Release gate table
One table per decided platform block, filled from that platform's checklist file.
| Column | Content |
|---|---|
| ID | The checklist item id, for example `I1` or `A2` |
| Grade | The grade copied from the checklist, unaltered: `[M]`, `[M-조건]`, `[G]` |
| Item | The obligation in one short sentence, with the threshold, measurement method, and percentile for a numeric item |
| Conclusion | `applies`, `not applicable` with the condition or reason, or `[확인 필요: <question>]` |
| Evidence | The document, console screen, or measurement that settles the conclusion |
| Re-check | The date the dated value is checked again, or `none` for an item with no dated value |
- Every `[M]` and `[M-조건]` item of the loaded checklist gets a row. A `[M-조건]` row whose condition is false reads "not applicable" with the reason; the row does not disappear.
- A `[G]` item gets a row only when it is relevant to this product.
- A `[M]` row left at `[확인 필요]` keeps the document at `status: draft`. That row is not counted toward the three-marker cap, which applies to inline markers in the prose sections.
- A store obligation is written as the clause plus the observable state the clause requires. "Follow the store guidelines" is not a conclusion.
## Data collection matrix
One matrix for the document, written only when the decided platforms include both `ios` and `android`.
| Column | Content |
|---|---|
| Data type | The data category, in the words the store forms use |
| Collected | `yes` or `no` |
| Shared with third parties | `yes` or `no`, with the recipient when `yes` |
| Linked to the user's identity | `yes` or `no` |
| Used for tracking | `yes` or `no` |
| Purpose | Why the release handles it |
| Optional | `yes` or `no`: whether the user may decline to provide the data and still use the feature |
| Ephemeral | `yes` or `no`: whether the data is processed only in transit and never stored |
- The matrix is written once. The App Privacy label and the Play Data safety declaration both read from it; neither restates the data types from memory.
- A type one platform handles and the other does not is a product decision: record the difference and its reason next to the matrix.
- A type the release handles but the matrix omits is a defect in the matrix, fixed before either store form is filled.
## Rules
- Copy the skeleton's section names exactly; they are the names a reviewer checks.
- The meta block carries the platform decision and the size decision, each with its one-line reason on the same line, for example `Platforms: web, ios - the request names a browser product and an App Store release`.
- Delete every placeholder you cannot fill with a fact, an assumption, or a marker. An empty section is a defect, not a template artifact.
- A `[가정: …]` that changes scope appears again as an `A-###` entry; a `[확인 필요: …]` that blocks scope becomes a `Q-###` with `Blocking: yes`.
- Every story gets one happy-path and one failure-or-edge scenario, so two stories produce four scenarios.
- Every decided platform block carries a release gate table. Every `[M]` and `[M-조건]` item of that platform's checklist gets a row with a conclusion; `[G]` items appear only when relevant.
- A numeric gate item states its threshold, its measurement method, and its percentile in the same cell.
- When `ios` and `android` are both decided, the data collection matrix is written once for the document, with the Apple privacy label view and the Play Data safety view derived from it.
- A store obligation is written as the clause plus the observable state the clause requires. A sentence that says to follow the guidelines without naming the clause and the state is not a requirement.
- At most three `[확인 필요: …]` markers in the whole document. This cap counts inline markers in the prose sections; a `[확인 필요: <question>]` that stands as a release gate table's conclusion is excluded from the count, and the document stays `status: draft` while any gate row is `[확인 필요]`.
- The handoff block is the only place the next stage's inputs are described; this document never contains a task list, a design, or a flow.
## Sources
> Checked 2026-11-07.
| Claim | Source |
|---|---|
| One document as the canonical artifact, with a short question budget | [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." |
| Removing a non-applicable section instead of writing N/A | [spec-kit specify](https://raw.githubusercontent.com/github/spec-kit/main/templates/commands/specify.md) - "When a section doesn't apply, remove it entirely (don't leave as 'N/A')." |
| Independently testable stories and numbered requirements | [spec-kit spec-template](https://raw.githubusercontent.com/github/spec-kit/main/templates/spec-template.md) - "Each user story/journey must be INDEPENDENTLY TESTABLE" |
| The WHEN/SHALL requirement sentence | [Kiro feature specs](https://kiro.dev/docs/specs/feature-specs.md) - "WHEN [condition/event] THE SYSTEM SHALL [expected behavior]" |
| State and unwanted-event requirement forms | [EARS](https://alistairmavin.com/ears/) |
| `SHALL CONTINUE TO` for unchanged behavior | [Kiro bugfix specs](https://kiro.dev/docs/specs/bugfix-specs.md) - "WHEN [condition] THEN the system SHALL CONTINUE TO [existing behavior]" |
| Inline markers at the gap | [deanpeters prd-development](https://github.com/deanpeters/Product-Manager-Skills) - "Tag inline where the gap appears, not just at the end." |
| Assumptions and open questions kept out of the body as fact | [phuryn create-prd](https://github.com/phuryn/pm-skills) - "Don't invent. What you don't know goes to 7.4 Assumptions and open questions, clearly marked, not into the body as fact." |
| Counter-metrics in the metric set | [TerminalSkills prd-writing](https://github.com/TerminalSkills/skills) - "Counter-metrics: [Metrics we don't want to hurt]" |
| A handoff block at the end of the document | [Citadel prd](https://github.com/SethGammon/Citadel) |
| One data inventory that both store privacy forms read from | [App Privacy Details](https://developer.apple.com/app-store/app-privacy-details/) - "This information is required to submit new apps and app updates to the App Store."; [Play Data safety](https://support.google.com/googleplay/android-developer/answer/10787469) - "Even developers with apps that do not collect any user data must complete this form" |
| No-naming of the internal build mechanism | [ai-conductor prd](https://github.com/jstoup111/ai-conductor) - "It MUST NOT name the new internal mechanism by which this feature is built." |
### Evidence grade
`LOCAL` - the section order and the worked example are this package's design. The sentence forms are quoted from the EARS and Kiro documentation; the marker and assumption rules are adapted from the cited guidance.
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: 321k | 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: 197k | 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: 197k | 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: 197k | Updated | Claude CodeCodex | Verified | Install |