Bezalel · Vitruvius · Rams · CRISP
(A design standard for AI agents)
AI has given every developer a design tool. It gave none of them a design standard.
CRISP gives your AI agent five clear tests for good design, so it can check its own work before you ship it.
(Install in 30 seconds)
The installer finds your AI coding tool and adds the skills to your project as plain files. You own them and can edit them.
Then run /crisp-teach once per project. It asks about your users, your design system and the products you admire, then writes .crisp.md. Almost every other command reads that file, so your agent remembers your project between sessions.
The files go where your tool expects them: .claude/ for Claude Code, .cursor/ for Cursor, .agents/skills/ for Antigravity. They only change when you choose to update.
(Why these skills exist)
You think your agent's design problem is prompt quality. It isn't. The agent has no way to judge its own output, so it ships the first thing that works. Every era of design solved this with a named standard. Ours had none. CRISP is that standard.
I · 1446 BC
"I have filled him with wisdom, with understanding, and with knowledge, and with all kinds of skill."
Exodus 31:3, on Bezalel, the first named designer
The oldest design standard on record is not about skill. Bezalel's third gift, da'at, means knowing who the work is for and why. An agent dropped into your codebase has none of it. So it builds for nobody in particular. Empty states say "No data available". Headings could belong to any product. One settings page serves the first-week beginner and the five-year expert alike.
The fix: give the agent that knowledge once. /crisp-teach asks you about your users, your design system and what good looks like for your product. It saves the answers in .crisp.md, so every later command starts with that context. /crisp-brief turns a vague request into a clear brief with success criteria you can check. /crisp-research shows what competitors do before you start designing.
II · 25 BC
"Well building hath three conditions: firmness, commodity, and delight."
Vitruvius, De Architectura
Code that compiles meets only the first condition, firmness (firmitas). Agents stop there. The interface works. But it shows a spinner every time you change a filter. It shows "1,247" with no comparison and no meaning. It sends you to another portal to approve one item. Technically correct. Hollow to use.
The fix: a real evaluation. /crisp-audit scores all five dimensions, rates each problem from P0 (critical) to P3 (minor) and lists the fixes in priority order. /crisp-review gives a grade and the top three issues in thirty seconds. /crisp-ux-laws and /crisp-a11y check the work against proven rules of how people think and the WCAG 2.2 AA accessibility standard.
III · 1976
"Good design is as little design as possible. Less, but better."
Dieter Rams
Left alone, an agent answers every request by adding something: another toggle, another card, another pop-up. Soon all twenty-three settings show to every user. Rams built his ten principles on the opposite idea: good design is about what you take away.
The fix: make removal a command. /crisp-redesign updates an existing UI without breaking what already works. /crisp-copy cuts interface text down to the words that help people find their way. /crisp-design-eng checks motion and polish, so nothing moves without a reason.
IV · 2026
"Your AI agent can generate code. It can't evaluate design."
CRISP
Close the session and the context is gone. The next session decides everything again: a new spacing scale, a different empty state, a fourth shade of grey. For three thousand years, people wrote design standards down so the knowledge would outlive the designer. Your agent's knowledge is gone in hours.
The fix: a written record the agent keeps up to date. .crisp.md keeps your context across sessions and updates. /handoff turns a reviewed design into a spec another agent can build from. /feature-design records why each flow decision was made. /crisp-ai applies the standard to AI features such as chat and agents, and /crisp-funnel builds a full funnel to the same standard.
Bezalel · Vitruvius · Rams · CRISP. 3,472 years of one question: does this design actually work for people?
(The framework)
CRISP gives your AI agent a shared design language. Each dimension names one common way designs fail, and how to fix it.
(Before & after)
AI agents build interfaces that work but feel wrong. Below, each dimension fails first. Then you see the same screen once it passes.
Can the user tell where they are and what this page does within 5 seconds?
No data available
Nothing in your Q3 pipeline yet.
Deals added here roll up to the forecast on Overview.
Add your first dealDoes the UI respond instantly to every action?
Does it show what the data means, not just the numbers?
Leads
1,247
Leads
1,247
23% vs last month3 deals need attention before Friday.
Does it fit into the user's day, or force them to fit into ours?
Expense approval required
£2,480 · Contractor invoice · Submitted by R. Achebe
You will need to sign in again.
Expense approval required
£2,480 · Contractor invoice · Submitted by R. Achebe
Already signed in. No redirect.
Does each user see only the complexity they need?
Illustrative interfaces. The names and figures in these examples are made up, not measured results.
Benchmarked against Stripe · Linear · Notion · Asana · Slack
(In session)
CRISP reads your project context, checks the work against the five dimensions and gives you a written verdict you can act on. Here is what a session looks like.
/crisp-review the settings page
Reading your project context first, then scoring the surface against all five dimensions.
/handoff once I have fixed those
Turning the reviewed design into a specification another agent can build from.
Run it once per project. It asks you a few questions, then writes the context file most other commands read.
What it returns
Illustrative output. Scores shown are an example, not a measured result.
The thorough check. It scores every dimension, names and rates every problem, and lists the fixes in the order to tackle them.
What it returns
Illustrative output. The grade shown is an example, not a measured result.
Thirty seconds, while you iterate. A grade and the three issues hurting users most, each with a fix.
What it returns
Describe the problem. Get user flows back, with the reason for each decision written down.
What it returns
Turns a reviewed design into a spec a build agent can follow without coming back to you with questions.
What it returns
(The skills)
All fourteen skills share one framework: the five dimensions. Most of them also read your project context from .crisp.md. Each one has its own page with the full details.
The core loop
Five commands that cover the whole design loop: set up, evaluate, design, hand off.
/crisp-teachRun once to set up. Learns your product and writes .crisp.md, the context file most other commands read.Returns.crisp.md context file · User personas and the jobs they need done · Design system reference · Products to learn from and products to avoiduser-invoked /crisp-auditA full check of all five dimensions. Rates every problem from P0 to P3 and gives a fix plan in priority order.ReturnsGrade A–F with justification · 5-dimension CRISP scorecard · Problems rated P0 (critical) to P3 (minor) · Prioritised fix listuser-invoked /crisp-reviewA thirty-second check. A grade from A to F and the three issues that hurt users most.ReturnsGrade A–F + one-line verdict · Top 3 issues ranked by impact · Specific fix per issue · Quick wins listuser-invoked /feature-designDescribe a problem. Get user flows, component choices and the reasons behind them.ReturnsUser flow with decision rationale · CRISP check at every step · Component recommendations · Open questions to test with usersuser-invoked /handoffTurns a reviewed design into a spec developers can build from.ReturnsComponent state inventory · Design token and spacing values · Edge cases and error states · Accessibility checklistuser-invokedSpecialist extensions
Deeper checks for research, accessibility, AI features, interface text, redesigns and the psychology of how people use screens.
/crisp-briefTurns a vague request into a clear brief with success criteria you can check.Returns.brief.md written to project root · Scoped problem statement · Observable success criteria · CRISP dimension priorityuser-invoked /crisp-copyChecks or writes interface text, such as buttons, errors and empty states, against the Contextual and Intelligent dimensions.ReturnsCopy audit with P0–P3 severity · Replacement strings ready to paste · Voice + tone consistency notes · Empty state copy (3-part structure)user-invoked /crisp-a11yA full WCAG 2.2 AA accessibility check, with exact code fixes.ReturnsWCAG 2.2 AA scorecard · Exact code fixes, not vague advice · a11y-checklist.md · Problems rated P0 (critical) to P3 (minor)user-invoked /crisp-aiChecks AI features, such as chat and agents, against six AI-specific dimensions.Returns6-dimension AI UI scorecard · Streaming UX recommendations · Patterns that show what the agent is doing · Error + uncertainty handlinguser-invoked /crisp-researchResearch before you design: what competitors do, what to avoid and which dimensions are at risk.ReturnsCompetitive pattern synthesis · CRISP dimension risk flags · Anti-patterns to avoid · Gaps in the brief, found before design startsuser-invoked /crisp-redesignRedesigns an existing UI without breaking what already works.ReturnsWhat to keep and what to rebuild · Upgrades in priority order · Current-state audit · Checks that SEO, analytics and accessibility still workuser-invoked /crisp-ux-lawsTurns proven laws about how people think into clear rules for each CRISP dimension.ReturnsEach law mapped to a CRISP dimension · Where interactions are too complex · One clear rule per law · Ways to cut choices and mental effortuser-invokedCraft and build
Motion and polish, plus the one skill that builds something instead of reviewing it.
/crisp-design-engMotion, micro-interactions, and the invisible details that make a UI feel right.ReturnsBefore and after table of craft problems · Motion decisions (duration · easing) · Performance + reduced-motion flags · Every fix mapped to a CRISP dimensionuser-invoked /crisp-funnelBuilds a complete mobile-first funnel from a brief: sections in order, copy written, work reviewed.ReturnsFunnel type and how much the buyer already knows · Sequenced sections (01 → 10) · Copy written to the offer · A /crisp-review check before handoveruser-invoked(Try it free)
Paste a URL and get a full CRISP audit of your product in seconds. It is free and needs no account.
(The CRISP letter)
Occasional emails on getting AI agents to produce work worth shipping. New skills are announced here first. No noise.
No spam. Unsubscribe any time.
(FAQ)
.crisp.md to your project root, and every CRISP command reads it automatically after that. Updating the skills never overwrites .crisp.md, so your context stays./crisp-audit and /crisp-review work on any screenshot or description, with no project context. /crisp-teach makes them much more accurate, because the agent then knows your users, your design system and what "good" means for your product.npx skills add @laith-wallace/crisp finds your AI coding tool and puts the files in the right folder: .claude/ for Claude Code, .cursor/ for Cursor, .agents/skills/ for Antigravity, and so on. If you prefer, download the universal ZIP and unzip it into your project root.Run a free audit on your product now. Or install the CRISP skills and give your AI agent a design standard.