nexu-io/open-design

reference-design-contract

Turn vague taste, screenshots, URLs, product notes, or "make it feel like this" references into a grounded DESIGN.md plus an implementation handoff. Use it before prototypes, decks, redesigns, or image remix work when the user needs a reusable visual direction rather than a one-off prompt.

77Collecting
See how to use itView GitHub source
npx skills add https://github.com/nexu-io/open-design --skill "skills/reference-design-contract"
Automated source guideDesignStandard source

Source checked Jul 28, 2026·Refresh due Oct 26, 2026

Reorganized from the pinned upstream SKILL.md

Source-grounded design guide: reference-design-contract

Use this skill when the user has taste signals, references, or a rough "like this" request and needs a reusable design contract before generation. The goal is not to write a longer prompt. The goal is to make design decisions explicit enough that a later prototype, deck, redesig…

npx skills add https://github.com/nexu-io/open-design --skill "skills/reference-design-contract"
Check the pinned source

The pinned source supports a structured brief, but not an expanded tutorial. Only detected inputs, outputs, and sections are shown.

476 source words · 3 usable sections

Design outputs

  • Make every claim traceable to user input, observed reference evidence, or an
  • Prefer concrete constraints over adjectives: "one warm accent, no purple or
  • Treat "do the same style" as "borrow controllable qualities", not "clone the

Design process

Read reference-design-contract through these 3 source sections

Sections are extracted automatically from the pinned SKILL.md and link back to the source.

01

Workflow

1. Lock the job. Identify target artifact type, audience, brand/product context, references, and constraints. Ask at most two questions only when a missing answer would change the direction. Otherwise choose a sensible default and label it as inferred. 2. Read evidence. Use prov…

SKILL.md · Workflow
Lock the job. Identify target artifact type, audience, brand/productRead evidence. Use provided screenshots, URLs, existing DESIGN.md,Separate reference semantics. For every reference, split it into:
02

What this skill produces

1. DESIGN.md — the reusable visual direction, following Open Design's standard nine-section design-system shape. 2. design-contract.md — the decision record: evidence used, keep/change boundaries, rationale, risks, and quality gate. 3. implementation-handoff.md — concise build i…

SKILL.md · What this skill produces
DESIGN.md — the reusable visual direction, following Open Design'sdesign-contract.md — the decision record: evidence used, keep/changeimplementation-handoff.md — concise build instructions for the next
03

Output rules

Make every claim traceable to user input, observed reference evidence, or an

SKILL.md · Output rules
Make every claim traceable to user input, observed reference evidence, or anPrefer concrete constraints over adjectives: "one warm accent, no purple orTreat "do the same style" as "borrow controllable qualities", not "clone the

Design checklist

Verify each item before delivery

The source section “Workflow” has been checked.

The source section “What this skill produces” has been checked.

The source section “Output rules” has been checked.

Source output checked: Make every claim traceable to user input, observed reference evidence, or an

Source output checked: Prefer concrete constraints over adjectives: "one warm accent, no purple or

Choose a different workflow

When another Skill is the better fit

design-review

Use when the user says "review the design", "check the UI", or wants a comprehensive UI/UX review. Uses a 7-phase methodology covering interaction, responsiveness, accessibility, and more.

A separate implementation from event4u-app/agent-config; compare its source, maintenance signals, and permission requirements.

Open source detail

higgsfield-generate

Use when the user wants to generate an image or video via Higgsfield AI. Covers 30+ models: Soul V2, Seedance 2.0, Kling 3.0, Veo 3.1, GPT Image 2, Nano Banana 2. Also covers Marketing Studio — branded ad video/image with avatars and products. Use whenever: "generate an image", "make a video", "animate this photo", "image-to-video", "img2vid", "edit this image with AI", "produce a clip", "create an ad", "make a UGC video", "marketing video", "brand video", "TV spot", "import product from URL", "

A separate implementation from MoizIbnYousaf/marketing-cli; compare its source, maintenance signals, and permission requirements.

Open source detail

arize-evaluator

Handles LLM-as-judge evaluation workflows on Arize including creating/updating evaluators, running evaluations on spans or experiments, managing tasks, trigger-run operations, column mapping, and continuous monitoring. Use when the user mentions create evaluator, LLM judge, hallucination, faithfulness, correctness, relevance, run eval, score spans, score experiment, trigger-run, column mapping, continuous monitoring, or improve evaluator prompt.

A separate implementation from github/awesome-copilot; compare its source, maintenance signals, and permission requirements.

Open source detail

FAQ

What does the reference-design-contract source document cover?

Use this skill when the user has taste signals, references, or a rough "like this" request and needs a reusable design contract before generation. The goal is not to write a longer prompt. The goal is to make design decisions explicit enough that a later prototype, deck, redesig…

How do I install reference-design-contract?

The source record exposes this install command: npx skills add https://github.com/nexu-io/open-design --skill "skills/reference-design-contract". Inspect the command and pinned source before running it.

Repository stars
82,073
Repository forks
9,485
Quality
77/100
Source repository last pushed

Quality breakdown

Based on traceable docs and repository signals; stars are not treated as quality.

77/100
Documentation20/30
Specificity20/25
Maintenance20/20
Trust signals17/25

Compare before choosing

Related Agent Skills and source variants

These links are selected from shared tasks, functions, stacks, platforms, and same-name variants. Compare the source owner, documentation, permissions, and maintenance signals.

design-review by event4u-app

Use when the user says "review the design", "check the UI", or wants a comprehensive UI/UX review. Uses a 7-phase methodology covering interaction, responsiveness, accessibility, and more.

higgsfield-generate by MoizIbnYousaf

Use when the user wants to generate an image or video via Higgsfield AI. Covers 30+ models: Soul V2, Seedance 2.0, Kling 3.0, Veo 3.1, GPT Image 2, Nano Banana 2. Also covers Marketing Studio — branded ad video/image with avatars and products. Use whenever: "generate an image", "make a video", "animate this photo", "image-to-video", "img2vid", "edit this image with AI", "produce a clip", "create an ad", "make a UGC video", "marketing video", "brand video", "TV spot", "import product from URL", "

arize-evaluator by github

Handles LLM-as-judge evaluation workflows on Arize including creating/updating evaluators, running evaluations on spans or experiments, managing tasks, trigger-run operations, column mapping, and continuous monitoring. Use when the user mentions create evaluator, LLM judge, hallucination, faithfulness, correctness, relevance, run eval, score spans, score experiment, trigger-run, column mapping, continuous monitoring, or improve evaluator prompt.

neuropixels-analysis by k-dense-ai

Analyze Neuropixels extracellular recordings end-to-end with SpikeInterface. Covers loading SpikeGLX/Open Ephys/NWB data, preprocessing, drift/motion correction, Kilosort4 (and CPU) spike sorting, quality metrics, and unit curation (threshold-based, model-based UnitRefine, and AI-assisted visual review). Use when working with Neuropixels 1.0/2.0 recordings, spike sorting, or extracellular electrophysiology analysis.

scientific-brainstorming by k-dense-ai

Facilitates evidence-aware scientific ideation with independent generation, structured discussion, explicit assumptions, transparent evaluation, adversarial review, and decision logs. Use for early-stage research brainstorming or prioritizing candidate directions; hand off empirical validation, study design, ethics or regulatory review, and clinical questions to appropriate experts or skills.

View original Skill.mdThis page is parsed directly from the repository SKILL.md without editorial rewriting. Collected: Jul 28, 2026 · about 3 min

Reference Design Contract

Use this skill when the user has taste signals, references, or a rough "like this" request and needs a reusable design contract before generation. The goal is not to write a longer prompt. The goal is to make design decisions explicit enough that a later prototype, deck, redesign, or image-remix run can execute without guessing.

What this skill produces

Create three files:

  1. DESIGN.md — the reusable visual direction, following Open Design's standard nine-section design-system shape.
  2. design-contract.md — the decision record: evidence used, keep/change boundaries, rationale, risks, and quality gate.
  3. implementation-handoff.md — concise build instructions for the next artifact-producing skill or coding agent.

If a preview is useful, also create example.html as a small hand-built contract preview. Do not make it the main deliverable.

Workflow

  1. Lock the job. Identify target artifact type, audience, brand/product context, references, and constraints. Ask at most two questions only when a missing answer would change the direction. Otherwise choose a sensible default and label it as inferred.
  2. Read evidence. Use provided screenshots, URLs, existing DESIGN.md, brand docs, image artifacts, or user notes. If evidence is missing, say so and base the contract on the brief only; do not invent brand facts.
  3. Separate reference semantics. For every reference, split it into:
    • Keep: qualities to preserve, such as density, composition, material, typography rhythm, color temperature, or motion attitude.
    • Change: subject matter, copy, brand marks, exact layout, protected assets, and anything the user wants adapted.
    • Do not copy: literal screenshots, logos, claims, pricing, proprietary UI, or exact prompt wording from examples.
  4. Freeze the direction. Choose one coherent visual stance. Do not provide five unrelated moodboards. If there are genuinely competing directions, name them briefly, pick the recommended one, and continue.
  5. Write DESIGN.md. Use these nine headings exactly:
    • ## 1. Visual Theme & Atmosphere
    • ## 2. Color
    • ## 3. Typography
    • ## 4. Spacing & Grid
    • ## 5. Layout & Composition
    • ## 6. Components
    • ## 7. Motion & Interaction
    • ## 8. Voice & Brand
    • ## 9. Anti-patterns
  6. Write design-contract.md. Include:
    • goal and target artifact
    • evidence table with confidence (observed, provided, inferred)
    • keep/change/do-not-copy table
    • final design stance in one paragraph
    • risks and explicit unknowns
    • quality gate checklist
  7. Write implementation-handoff.md. Keep it short and operational:
    • files to read
    • token/palette/type/layout constraints
    • asset rules
    • responsive requirements
    • "first artifact should prove..." acceptance notes
  8. Validate. Read references/checklist.md and satisfy every P0 gate before final handoff.

Output rules

  • Make every claim traceable to user input, observed reference evidence, or an explicitly labeled inference.
  • Prefer concrete constraints over adjectives: "one warm accent, no purple or blue glow" beats "premium".
  • Treat "do the same style" as "borrow controllable qualities", not "clone the original subject or prompt".
  • If the user asks for immediate UI generation too, finish these contract files first, then hand off to the appropriate artifact skill in the next step.
Skill path
skills/reference-design-contract/SKILL.md
Commit SHA
89d6d4ef21ba
Repository license
Apache-2.0
Data collected