agentleFS
sreejithramansreejithraman/SreeStack/skills/prototype/SKILL.md

prototype

Skill1 starsChanged 16 days agoMITView on GitHub

Build runnable, disposable UI prototypes for web, mobile, or desktop. Use to answer a design question by trying one concept or comparing meaningfully different layouts or interactions before production implementation.

194 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use playtest-report, soak-test and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • CoralGameSkill

    playtest-report

    Generates a structured playtest report template or analyzes existing playtest notes into a structured format. Use this to standardize playtest feedback collection and analysis.

    Also in 12 projects that use this

  • lilangMaxSkill

    soak-test

    Generate a soak test protocol for extended play sessions. Defines what to observe, measure, and log during long play sessions to surface slow leaks, fatigue effects, and edge cases that only appear after sustained play. Primarily used in Polish and Release phases.

    Also in 9 projects that use this

  • lilangMaxSkill

    team-live-ops

    Orchestrate the live-ops team for post-launch content planning: coordinates live-ops-designer, economy-designer, analytics-engineer, community-manager, writer, and narrative-director to design and plan a season, event, or live content update.

    Also in 9 projects that use this

  • CoralGameSkill

    team-polish

    Orchestrate the polish team: coordinates performance-analyst, technical-artist, sound-designer, and qa-tester to optimize, polish, and harden a feature or area for release quality.

    Also in 14 projects that use this

  • +5 more from lilangMax
  • +3 more from CoralGame

The file

Show its rules and text

We split this skill into the rules it gives an agent

24 rules. 0 are common, 20 are distinctive, and 4 are very specific to this project.

  • 0

    Common

    Most of the 179 edited versions we read keep these.

  • 20

    Distinctive

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use one concept when the question is whether a specific direction works.

    line 20
  2. Distinctive:

    Use two when the decision is a genuine tradeoff.

    line 21
  3. Distinctive:

    Add further options when each represents a distinct direction that the existing set does not test.

    line 22
  4. Distinctive:

    Prefer the existing screen with its surrounding navigation, components, and representative content.

    line 30
  5. Very specific:

    Use ui-design for visual direction and for browser behavior or accessibility.

    line 37
  6. Very specific:

    Use animate when motion is part of the question.

    line 39
  7. Very specific:

    Use swiftui or uikit for native Apple construction and behavior.

    line 40
  8. Distinctive:

    Web: use the existing route when possible.

    line 44
  9. Distinctive:

    SwiftUI or UIKit: use the app's framework. A preview can answer a static layout question.

    line 46
  10. Distinctive:

    Other mobile or desktop platforms: use the app's normal framework and host, with the relevant input methods and window sizes.

    line 49
  11. Distinctive:

    Use a browser mockup for a native product only when the user requests or accepts that fidelity.

    line 52
  12. Distinctive:

    Use the project's components and design system while allowing each option to make its intended structural choice.

    line 58
  13. Distinctive:

    Use the same representative content and meaningful empty, loading, error, or crowded states across options.

    line 62
  14. Distinctive:

    Add only enough behavior to try the flow.

    line 65
  15. Distinctive:

    When the requested workflow would write real data or cause another external effect, keep that step stubbed, state exactly what was not …

    line 67
  16. Distinctive:

    Keep the code disposable but reliable enough to evaluate.

    line 74
  17. Distinctive:

    Mobile: use a debug menu, sheet, or compact native picker that respects safe areas and existing gestures.

    line 86
  18. Distinctive:

    Desktop: use a development menu or toolbar suited to the app.

    line 88
  19. Distinctive:

    Keep sample state stable while switching when that makes the comparison fair.

    line 90
  20. Distinctive:

    When options require different state or navigation, reset to a known starting point and make the reset visible.

    line 91
  21. Very specific:

    Run every option on the target platform and exercise the interactions that answer the design question.

    line 97
  22. Distinctive:

    Leave the choice open until the user picks one or asks you to choose.

    line 109
  23. Distinctive:

    Before a decision, keep the prototype reviewable on the task branch or another clearly marked development surface.

    line 114
  24. Distinctive:

    When production implementation is in scope, apply the chosen direction with the project's normal quality checks and remove prototype …

    line 119

“Common” is measured: kept by at least 6 in 10 of the 179 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Prototype

Build enough UI to answer a design question on the target platform. A prototype is an exploration surface, not authorization to ship the result.

1. Frame the experiment

Inspect the request, current interface, nearby code, and available tooling. State the design question, where the prototype will run, and what evidence would make the options distinguishable. Ask only when a missing choice would materially change what you build; continue safe, reversible work while waiting.

Choose an option set large enough to expose the real decision:

  • Use one concept when the question is whether a specific direction works.
  • Use two when the decision is a genuine tradeoff.
  • Add further options when each represents a distinct direction that the existing set does not test.

Variants should differ in layout, information hierarchy, navigation, behavior, or motion rather than only color or copy. Follow a count the user specifies.

2. Choose the real surface

Prefer the existing screen with its surrounding navigation, components, and representative content. Use a separate prototype route, screen, preview, or development target when no existing screen fits. Clearly mark prototype files, entry points, and controls, and keep them out of release builds.

Load only the specialists the experiment needs:

  • Use ui-design for visual direction and for browser behavior or accessibility.
  • Use animate when motion is part of the question.
  • Use swiftui or uikit for native Apple construction and behavior.

Match the platform:

  • Web: use the existing route when possible. A query parameter can make each option easy to reopen and share.
  • SwiftUI or UIKit: use the app's framework. A preview can answer a static layout question; use the running app for navigation, gestures, keyboard behavior, system UI, or other runtime behavior.
  • Other mobile or desktop platforms: use the app's normal framework and host, with the relevant input methods and window sizes.

Use a browser mockup for a native product only when the user requests or accepts that fidelity. When target tooling is unavailable, state the blocking tooling gap and what the fallback can and cannot establish.

3. Build safely

Use the project's components and design system while allowing each option to make its intended structural choice. Share stable fixtures and controls rather than forcing the options through one premature abstraction.

Use the same representative content and meaningful empty, loading, error, or crowded states across options. Reuse safe read-only data access when helpful. Stub writes, keep experimental state local, and prevent external side effects. Add only enough behavior to try the flow.

When the requested workflow would write real data or cause another external effect, keep that step stubbed, state exactly what was not exercised, and offer a provider test mode, disposable account, or local fixture instead.

Spend effort on the dimensions the question tests, such as hierarchy, spacing, touch targets, safe areas, keyboard overlap, resizing, or motion character.

Keep the code disposable but reliable enough to evaluate. Prefer running the prototype over production-grade automated coverage. Add a narrow test only when non-obvious experimental logic would otherwise make the result untrustworthy.

4. Make comparison easy

For multiple options, provide one development-only control that identifies the current option and switches without rebuilding. Keep it visually separate from the interface under review and away from its controls.

  • Web: a switcher can update the option through the project's router. Keyboard shortcuts must leave text inputs and other controls alone.
  • Mobile: use a debug menu, sheet, or compact native picker that respects safe areas and existing gestures.
  • Desktop: use a development menu or toolbar suited to the app.

Keep sample state stable while switching when that makes the comparison fair. When options require different state or navigation, reset to a known starting point and make the reset visible. Give each option a short name that states its design choice.

5. Exercise and present

Run every option on the target platform and exercise the interactions that answer the design question. Use verify when the judgment depends on a real workflow. Use showroom to choose demonstrations and access.

For multiple options, exercise the switcher and confirm that its control stays clear of the interface and interactions under review.

Provide the run command or build target, entry screen, option controls, and any environment prerequisite needed to return to the prototype. Identify each screenshot or recording by option. Explain the tradeoff each option tests and state any behavior or platform condition that remains unverified.

Leave the choice open until the user picks one or asks you to choose. Combine parts only when the resulting direction remains coherent and testable.

6. Record and dispose

Before a decision, keep the prototype reviewable on the task branch or another clearly marked development surface. Once a direction is chosen, record the winner and why in a durable project surface already in scope, such as a PR description, project note, or authorized issue update.

When production implementation is in scope, apply the chosen direction with the project's normal quality checks and remove prototype controls and unused options. Preserve the prototype branch or artifacts when the user requests it, revisiting the comparison still has value, or no durable project record is authorized. When retained, record the branch pointer and run steps in the durable record when one is available. Otherwise retain the decision and evidence there, then clean up the disposable work.

A prototype-only request ends with a reviewable prototype, not a production change.

Done

The prototype is ready for a decision when the design question is explicit, each option tests a distinct answer, relevant states and interactions run on the target surface, comparison is easy, and the evidence and remaining gaps are clear. After a decision, completion also requires a recorded verdict and an explicit disposition for the disposable work.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.