ckm:design

Overview of ckm:design skill

What ckm:design skill actually does

ckm:design is a broad design production skill for people who want an AI agent to do more than generate a vague “make it look better” answer. It bundles practical workflows for logo design, corporate identity program (CIP) mockups, banners, slide decks, icons, social photos, and design routing to other specialized skills such as brand, design-system, and UI styling.

Who should install ckm:design skill

This skill is best for users who need structured design output fast: founders creating a first brand package, marketers producing platform-specific assets, product teams exploring visual directions, and agents that need references, CSV-backed options, and runnable scripts instead of generic design advice.

Best-fit jobs-to-be-done

Use ckm:design when your real goal is one of these:

• turn a rough brand idea into a logo direction
• create CIP deliverables and mockup concepts
• generate banners sized for real channels
• build HTML slides with layout guidance
• create SVG icons with a chosen visual style
• route a request to the right sub-skill instead of forcing one prompt to do everything

Why this skill is more useful than a plain prompt

The main differentiator is structure. The repository includes:

• routing guidance in references/design-routing.md
• concrete size/style references for banners, slides, logo, CIP, and icons
• CSV-backed search data for industries, styles, colors, and deliverables
• Python scripts for CIP search, CIP generation, HTML rendering, and icon generation

That means ckm:design skill is stronger when you want repeatable output or need the agent to choose from known design options rather than invent everything ad hoc.

What to know before adopting

This is not a single “design everything automatically” command. It is a hub skill with several built-in tracks. Adoption is easiest if you already know whether you need logo, CIP, slides, banner, or icon work. If your need is specifically design tokens, shadcn/ui implementation, or brand strategy, the routing docs explicitly point you to adjacent skills instead of overloading ckm:design.

How to Use ckm:design skill

Install context for ckm:design install

The repository does not expose a built-in install command inside SKILL.md, so the practical pattern is to add the source repo as a skill and target design:

npx skills add https://github.com/nextlevelbuilder/ui-ux-pro-max-skill --skill design

After install, the local path is typically under ~/.claude/skills/design, which matters because several documented commands assume that location.

Read these files first

For the fastest understanding of ckm:design usage, start in this order:

1. SKILL.md
2. references/design-routing.md
3. the reference file for your task:
   • references/logo-design.md
   • references/cip-design.md
   • references/slides.md
   • references/banner-sizes-and-styles.md
   • references/icon-design.md

If you are evaluating output quality before committing, also inspect:

• references/cip-style-guide.md
• references/cip-prompt-engineering.md
• references/logo-prompt-engineering.md
• data/ CSV files for available styles, industries, and options

Start by routing the task correctly

A common failure with ckm:design for UI Design is using it for work the repo intends to route elsewhere. Use references/design-routing.md to decide:

• use brand for brand identity and voice
• use design-system for tokens and CSS variables
• use ui-styling for shadcn/ui, Tailwind, and implementation
• use built-in design tracks for logo, CIP, slides, banners, and icons

This routing step matters because the repository is opinionated: better results come from the specialized path, not from asking the top-level design skill to improvise.

Inputs that produce better output

Before calling ckm:design skill, prepare these inputs:

• deliverable type: logo, business card mockup, LinkedIn banner, pitch deck, icon set
• industry or product category
• brand name
• audience
• preferred style adjectives
• required platform or size constraints
• color preferences or forbidden colors
• examples of what “good” looks like
• final output format needed: prompt, SVG, HTML, mockup concept, asset list

Strong inputs reduce drift. “Design a banner for our startup” is weak. “Create a LinkedIn company cover for a B2B AI compliance startup, clean enterprise style, navy/white palette, leave safe headline zone, optimize for credibility not hype” is much more aligned with the repository’s structure.

How to turn a rough goal into a usable prompt

A good prompt for ckm:design guide usually has five parts:

1. objective
2. asset type
3. brand and audience context
4. style and constraints
5. desired output shape

Example:

“Use ckm:design to create a concept for a consulting brand refresh. I need a primary logo direction plus 3 CIP deliverables to mock up first. Brand: NorthPeak Advisory. Audience: mid-market CFOs. Style: corporate minimal, trustworthy, not old-fashioned. Colors: navy, white, restrained accent color. Deliverables: business card, letterhead, office signage. Output: recommended style, rationale, and mockup prompts.”

That works because it maps directly to the repo’s style guides, deliverable lists, and prompt-engineering references.

Using the CIP workflow well

The CIP track is one of the most operationally useful parts of ckm:design. It combines reference docs with scripts and searchable data.

A practical start:

• read references/cip-design.md
• read references/cip-deliverable-guide.md
• generate a brief before generating images

Example commands from the repo:

python3 ~/.claude/skills/design/scripts/cip/search.py "tech startup" --cip-brief -b "BrandName"

python3 ~/.claude/skills/design/scripts/cip/search.py "business card letterhead" --domain deliverable

This is better than freeform prompting because it gives the agent real deliverable and style anchors before mockup generation.

When to use the CIP scripts instead of chat only

Use the scripts when you need:

• searchable deliverables, industries, or styles
• repeatable mockup generation across several assets
• HTML rendering from generated mockups
• a more deterministic workflow than chat memory alone

If you just need brainstorming, chat is enough. If you need a consistent package with named deliverables, the scripts materially improve ckm:design usage.

Using ckm:design skill for banners

The banner workflow is especially useful when size mistakes would waste time. references/banner-sizes-and-styles.md contains real platform dimensions and art-direction styles across social, ads, web, and print.

Good banner requests should include:

• target platform
• exact asset type
• CTA priority
• safe-area concerns
• whether the design is conversion-first or brand-first

Example:
“Use ckm:design to plan a YouTube channel art banner for a developer tools brand. Respect safe area, keep headline centered, minimal style, dark theme, emphasize product clarity over decoration.”

Using ckm:design skill for icons

For icon work, the repo is more than a style list: it includes a generator script and supports batch and size variants.

Useful commands include:

• python3 ~/.claude/skills/design/scripts/icon/generate.py --prompt "settings gear" --style outlined
• python3 ~/.claude/skills/design/scripts/icon/generate.py --prompt "user profile" --sizes "16,24,32,48" --output-dir ./icons

This makes ckm:design install more attractive if you need production-oriented icon exploration rather than a one-off icon prompt.

Using ckm:design skill for slides

The slides references are worth using when you need structured presentations, not just copy bullets. The repository includes:

• references/slides.md
• references/slides-create.md
• references/slides-html-template.md
• references/slides-layout-patterns.md
• references/slides-strategies.md
• references/slides-copywriting-formulas.md

This is useful when your agent needs to convert a narrative into HTML-based slides with layout patterns and presentation strategy, not just draft text.

Practical workflow that works in real projects

A reliable workflow for ckm:design:

1. classify the design task with references/design-routing.md
2. read the task-specific reference
3. gather missing constraints from the user
4. use the relevant data or script if the track supports it
5. generate one strong direction first, not five shallow ones
6. refine based on platform, style, and audience feedback
7. only then expand into variations or multi-asset sets

This avoids the biggest adoption problem: producing lots of output before the style direction is validated.

ckm:design skill FAQ

Is ckm:design skill good for beginners?

Yes, if the beginner already knows the kind of asset they need. The references are practical and concrete. The main challenge is scope: because ckm:design skill covers many design tasks, beginners should start with one track such as logo, banner, or CIP rather than the entire skill at once.

Is ckm:design for UI Design the right choice?

Only sometimes. If you need visual direction, brand-adjacent styling, or high-level design guidance, it can help. If you need actual component implementation, Tailwind classes, token architecture, or shadcn/ui work, the repo’s own routing guidance says to use the dedicated ui-styling or design-system skills instead.

What makes ckm:design better than a normal design prompt?

The value is not magic image quality by itself. It is the repository’s operating scaffolding: styles, dimensions, CSV data, prompt structures, and scripts. That lowers guesswork and gives the agent more grounded options than a blank-prompt workflow.

Does ckm:design install require extra tools?

Potentially yes for script-based workflows. The references document Python commands under ~/.claude/skills/design/scripts/.... Some flows also mention Gemini-based generation. If you only want chat guidance, you can still benefit without running scripts, but some of the strongest workflows depend on your local environment supporting those commands.

When is ckm:design skill a bad fit?

Skip it if:

• you only need a tiny one-off visual suggestion
• you want pixel-perfect UI implementation in code
• you need a fully integrated design app, not agent-guided references and scripts
• you are unwilling to provide constraints like platform, style, or audience

Can it generate complete brand systems by itself?

Not cleanly from the top-level skill alone. The design repo intentionally routes broader brand identity and design-system tasks to neighboring skills. ckm:design guide is best seen as a practical design execution hub, not the sole source of truth for every branding and UI decision.

How to Improve ckm:design skill

Improve ckm:design results by narrowing the asset first

The fastest quality gain is specificity. Ask for “a LinkedIn company cover” or “three CIP deliverables for a legal consultancy,” not “some branding.” The repository is strongest when the asset class is explicit.

Give style inputs that match the repo vocabulary

Use style language the references already support, such as:

• corporate minimal
• modern tech
• luxury premium
• warm organic
• bold dynamic

This helps the agent align with the actual style guides in references/cip-style-guide.md instead of inventing a new taxonomy mid-task.

Include platform and format constraints early

For banners, specify the exact platform and size-sensitive area. For icons, specify sizes and stroke expectations. For slides, specify whether the output should be narrative, investor-facing, educational, or sales-oriented. These details change output quality more than adding more adjectives.

Avoid the most common failure mode

The biggest failure mode in ckm:design skill is mixing multiple jobs in one prompt: logo + website UI + design tokens + ad banners + pitch deck. That produces generic output. Split the workflow:

• validate visual direction
• generate one asset class
• expand only after approval

Use the data and scripts when available

If a track has CSV data or scripts, use them. For example:

• CIP: search before generation
• icons: list styles before choosing one
• banners: use documented dimensions rather than “standard size”
• slides: follow the HTML/template references before drafting layouts

This is the practical difference between average and strong ckm:design usage.

Ask for rationale, not just assets

A better prompt asks for:

• selected style
• why it fits the audience
• what constraints drove the choice
• what to test next

That makes iteration easier. If the first output misses the mark, you can adjust the style logic instead of restarting from scratch.

Iterate with contrast, not tiny tweaks

After the first result, request 2 to 3 meaningfully different directions:

• conservative vs bold
• premium vs approachable
• enterprise vs creator-friendly

That is more useful than asking for ten small variations, especially in brand and banner tasks.

Improve ckm:design for team handoff

If the output will go to a designer, marketer, or developer, ask the skill to package decisions clearly:

• dimensions
• color palette
• typography direction
• style name
• deliverable list
• export needs
• prompt or script used

That turns ckm:design from a brainstorming tool into a handoff tool.

Repository areas worth exploring next

If you adopt this skill, the highest-value deeper reads are:

• references/design-routing.md
• references/cip-design.md
• references/cip-prompt-engineering.md
• references/banner-sizes-and-styles.md
• references/icon-design.md
• data/cip/ and data/logo/ CSVs for supported option spaces

These files reveal how much structure the skill actually gives you, which is the core install decision.