agentleFS
dafaalzdafaalz/agent-skills/skills/prototype/SKILL.md

prototype

Skill2 starsChanged 51 days agoMITView on GitHub

Use when running live interactive prototyping sessions, generating design variants, rapid mockups, or establishing browser sandbox loops. Don't use for production components (use frontend-design), UX audits (use ui-ux-review), CSS animations (use ui-motion), React profiling (use react), or multi-task execution plans (use agent-planning).

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

13 rules. 0 are common, 8 are distinctive, and 5 are very specific to this project.

  • 0

    Common

    Most of the 179 edited versions we read keep these.

  • 8

    Distinctive

    This author's own choices. Worth a look.

  • 5

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Write-Target Boundary. In Exploration branch, write only under .agents/scratch/prototypes/<variant-id>/ and the isolated preview route path …

    line 13
  2. Distinctive:

    Keep interaction transitions under 300ms, animate transform and opacity exclusively, use ease-out curves on entrances, and respect …

    line 27
  3. Distinctive:

    Delete the prototype harness and unselected variants once the user confirms a winning direction.

    line 30
  4. Distinctive:

    Follow these six steps in sequence:

    line 43
  5. Very specific:

    If .agents/scratch/prototypes/<variant-id>/spike-brief.md exists, read it as the authoritative brief.

    line 47
  6. Distinctive:

    If the file does not exist, isolate the target UI component from the user request as described below.

    line 49
  7. Distinctive:

    If the request targets a complex screen or multi-component flow, select the component that carries the highest interaction risk or design …

    line 50
  8. Distinctive:

    Verify that no two variants occupy the same point on a chosen axis.

    line 72
  9. Very specific:

    When the user selects a winner, write .agents/scratch/prototypes/<variant-id>/PROTOTYPE_LOCK.md containing:

    line 106
  10. Very specific:

    Delete the prototype route, variant files, and harness unless the user explicitly requests retention.

    line 114
  11. Very specific:

    Keep prototype code strictly confined to scratch preview paths (/prototypes/ or temporary HTML files):

    line 120
  12. Distinctive:

    Keep prototype edits confined to scratch preview paths.

    line 121
  13. Distinctive:

    Run this check before completing any prototyping pass:

    line 136

“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

Prototyping variants

Explore divergent visual and interaction directions for a single UI component behind an interactive picker harness.

This skill builds distinct, production-ready variants for a single UI component and renders them in an isolated picker harness. The user evaluates variants in context and selects a direction to integrate.

<HARD-GATE> Write-Target Boundary. In Exploration branch, write only under `.agents/scratch/prototypes/<variant-id>/` and the isolated preview route path declared in `.agents/state/prototype/harness-path.txt`. Never modify production component files, routing tables, or global build configuration. In Promotion branch, migrate only the winning variant into production after user confirmation recorded in `.agents/scratch/prototypes/<variant-id>/PROTOTYPE_LOCK.md`. </HARD-GATE>

Modes

Select an execution branch based on user invocation:

  • Exploration branch (default). Run Steps 1 through 5 to research, build, and verify candidate variants.
  • Promotion branch (keep <variant>). Skip directly to Step 6 to integrate the chosen component and clean up scratch files.

Core rules

  • Build all variants and the picker harness in an isolated route or standalone HTML file instead of editing production components during exploration.
  • Differentiate variants by layout, density, motion, or interaction pattern rather than presenting variants that vary only by color tint or placeholder text.
  • Supply realistic domain copy, interactive states, and transitions for every variant instead of relying on placeholder text or dead buttons.
  • Keep interaction transitions under 300ms, animate transform and opacity exclusively, use ease-out curves on entrances, and respect reduced-motion preferences.
  • Render the picker component exactly as defined in PICKER.md instead of restyling it with project tokens or theme colors.
  • Emit a viewable v0 draft containing layout skeletons, design tokens, and explicit module placeholders (such as [image] or [icon]) to confirm direction with the user before investing in complex interaction choreography or micro-motion.
  • Delete the prototype harness and unselected variants once the user confirms a winning direction.

Reference quick routing

Workflow domainReference moduleScope
Live session loopreferences/live-session-loop.mdLive browser prototyping session control
Variant generationreferences/variant-generation.mdGenerative design permutations & A/B variants
Live generate workflowreferences/live-generate-workflow.mdGeneration pipelines & artifact output
Live setup guidereferences/live-setup-guide.mdFirst-time configuration & server prerequisites

Workflow

Follow these six steps in sequence:

Step 1. Scope and brief definition

If .agents/scratch/prototypes/<variant-id>/spike-brief.md exists, read it as the authoritative brief. Extract target_component, target_surface, and primary_job from the file. Skip the remaining Step 1 isolation instructions unless the brief is missing or malformed.

If the file does not exist, isolate the target UI component from the user request as described below.

  • If the request targets a complex screen or multi-component flow, select the component that carries the highest interaction risk or design ambiguity.
  • State the brief in one declarative sentence defining the component, its target surface, and its primary job.

Completion criterion. .agents/scratch/prototypes/<variant-id>/brief.md exists on disk with a single # Brief heading and one sentence under it. If spike-brief.md was read, the brief content matches the values in that file.

Step 2. Technical and visual recon

Inspect the active codebase to identify design constraints and styling tokens.

  • Stack. Identify the UI framework, styling library, and animation libraries in use.
  • Tokens. Extract active values for spacing, border radii, type scale, and core palette.
  • Context. Inspect the container dimensions, background surfaces, and sibling components where the target piece mounts.
  • Standalone fallback. If no existing codebase exists, default to semantic HTML, standard system fonts, neutral slate tones, and one accent color.

Completion criterion. .agents/scratch/prototypes/<variant-id>/inventory.json exists on disk recording target stack, design tokens, and surrounding DOM constraints.

Step 3. Direction and axis selection

Define three distinct directions before writing code. Expand to five only when the design space warrants deeper exploration.

  • Assign each direction a descriptive name such as Quiet, Editorial, Playful, or Dense.
  • Pair each direction with an explicit axis of variation, such as density, spatial hierarchy, interaction cadence, or feedback mechanism.
  • Verify that no two variants occupy the same point on a chosen axis.

Completion criterion. .agents/scratch/prototypes/<variant-id>/directions.json exists on disk specifying 3 to 5 named directions with mutually exclusive variation axes.

Step 4. Harness construction

Assemble the isolated preview environment and embed the standard picker.

  • Dev server environment. Create a dedicated route or preview page at /prototypes/<component-slug> containing variant components and the harness.
  • Static environment. Generate a single self-contained HTML file containing inline styles and scripts for direct browser execution.
  • Picker integration. Copy markup, styles, and event wiring verbatim from PICKER.md.
  • Display standards. Render one variant at a time at full size with surrounding layout context instead of using thumbnail grids, and switch variants instantly without transitional animations.

Completion criterion. The operational harness renders variants one at a time and switches instantly via keyboard and mouse controls.

Step 5. Verification and review handoff

Validate variant functionality before presenting options to the user.

  • Flip through every variant in the harness to verify state transitions and event responses.
  • Inspect the browser console to confirm zero runtime errors or unhandled exceptions.
  • Present the comparison table summarizing trade-offs:
#VariantAxisBest suited forTrade-off
1QuietMinimal motion, borders over shadowsHigh-frequency daily toolsLowest visual impact
2EditorialHigh contrast typography, generous whitespaceOnboarding or milestone momentsConsumes vertical space
  • Report the active harness URL or local file path along with keyboard shortcut instructions.

Completion criterion. The harness executes cleanly without console warnings, and the comparison table articulates distinct trade-offs for each variant.

Step 6. Variant promotion and cleanup

  1. When the user selects a winner, write .agents/scratch/prototypes/<variant-id>/PROTOTYPE_LOCK.md containing:
    • variant_id
    • selected_direction (the direction name from Step 3)
    • selection_timestamp (ISO 8601)
    • harness_path (from .agents/state/prototype/harness-path.txt)
    • files_to_promote
  2. Copy the lock file to .agents/state/prototype/locks/<variant-id>.md for durable record.
  3. Migrate the winning component into production paths following codebase file structure and token conventions.
  4. Delete the prototype route, variant files, and harness unless the user explicitly requests retention. Do not delete PROTOTYPE_LOCK.md or its archived copy.

Completion criterion. .agents/scratch/prototypes/<variant-id>/PROTOTYPE_LOCK.md exists on disk with all five fields. .agents/state/prototype/locks/<variant-id>.md exists with matching content. ls /prototypes/ (or the framework equivalent path from harness-path.txt) returns zero prototype files unless the user requested retention.

Boundary isolation

Keep prototype code strictly confined to scratch preview paths (/prototypes/ or temporary HTML files):

  • Keep prototype edits confined to scratch preview paths; never modify production component files, routing tables, or global build configurations during exploration.
  • Strip prototype harnesses and scratch dependencies before merging; do not commit prototype harnesses or scratch dependencies to production bundles.

Invocation variants

InvocationBehavior
<description>Full workflow covering scope, recon, 3 variants, picker, and waiting for choice
<description> x5Full workflow with 5 variants
riff <variant>New round keeping the harness and diverging around the chosen variant
keep <variant>Promote that variant into the codebase and delete the prototype surface
keep <variant>, leave the pickerPromote that variant while retaining the prototype surface

Quick audit checklist

Run this check before completing any prototyping pass:

CheckPassing condition
FrontmatterDescription starts with "Use when" and lists operational triggers under 500 characters
IsolationZero prototype imports or scratch dependencies exist inside production files
Functional parityEvery variant executes real interactions without placeholder buttons or lorem ipsum text
Motion limitsAll UI transitions run under 300ms using transform and opacity properties
Picker fidelityPicker markup, styles, and keyboard handlers match PICKER.md verbatim
Completion criteriaEvery workflow step defines a checkable binary completion criterion
Style complianceZero em dashes, zero decorative emojis, and sentence case headings throughout

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.