An AI agent skill that takes you from research insight to design specification — grounded in industry-standard methodology.

One plain-language prompt in, a research-backed deliverable out. Two modules: UX Research and UI Design.

---

See It In Action

Example output

Persona	Journey Map	Design Tokens	Canvas Artifact

---

Built by Ravi Palwe — 19+ years in Product Design leadership in Financial Services. PRISM encodes that methodology into a skill, so the output reflects how a senior practitioner actually works, not a generic prompt.

What It Does

Module 1: UX Research

Deliverable	What You Get
User Personas	3–5 research-backed personas with goals, behaviors, pain points, motivations, context of use, and design implications
Journey Maps	Phase-by-phase experience maps with emotional arcs, touchpoints, and prioritized opportunities
Interview Synthesis	Cross-participant findings with evidence strength ratings, direct quotes, themes, and recommendations
Usability Test Plans	Complete plans with research questions, task scenarios, metrics, participant criteria, and timelines

Module 2: UI Design

Deliverable	What You Get
Design Philosophy	Named aesthetic direction with vision, principles, emotional targets, and anti-patterns
Design Tokens	Complete color, typography, spacing, elevation, radius, and motion token system
Component Specs	Anatomy, states, behavior, responsiveness, accessibility, and Do/Don't for every component
Layout System	Grid, breakpoints, page templates, content regions, navigation patterns, and empty states
Interaction Spec	Motion principles, timing system, feedback patterns, transitions, gestures, and loading states
Canvas Artifacts	Visual design expressions as .pdf or .png — posters, infographics, brand identity pieces

Financial Services Layer (Optional)

Built-in support for fintech, banking, payments, insurance, and wealth management. Adds trust signals, compliance friction (KYC, AML, PCI-DSS), financial anxiety analysis, trust-specific design tokens, financial component patterns, and regulatory touchpoints to every deliverable.

Install

Automated (Recommended)

npx skills add ravipalwe/prism

Manual

Add the path to SKILL.md to your .claude/settings.json:

{
  "skills": ["path/to/prism/SKILL.md"]
}

Quick Start

Install, then just describe what you need in plain language:

Create user personas for a mobile banking app targeting Gen Z users.

PRISM identifies the module and deliverable, classifies your input fidelity, and follows the relevant template.

Usage Examples

Research:

• "Create user personas for a mobile banking app targeting Gen Z users"
• "I have interview notes from 6 users about our onboarding flow — synthesize them"
• "Write a usability test plan for our new peer-to-peer payments feature"
• "Map the customer journey for opening a brokerage account, include the financial services layer"

Design:

• "Create a design system for a mobile banking app — calm, trustworthy, professional"
• "I need component specs for a payment confirmation flow"
• "Design tokens for a wealth management dashboard — include the financial services layer"
• "Generate a layout system for an insurance comparison app"

Cross-Module Flow

Research deliverables feed design deliverables:

Research Output	Feeds Into
Personas	Emotional targets in design philosophy, accessibility personas
Journey Maps	Key moment identification, state design, flow prioritization
Interview Synthesis	Pain point → design solution mapping
Usability Test Plans	Component testing criteria, interaction success metrics

Benchmark Results

Tested across 3 eval scenarios (personas, interview synthesis, usability test plan):

Metric	With PRISM	Without PRISM	Delta
Pass Rate	100%	36%	+64%
Avg Tokens	55K	47K	+8K
Avg Time	337s	304s	+33s

PRISM adds modest cost (~33 seconds, ~8K tokens) for a 64% quality improvement on research methodology compliance, evidence-based structure, and actionable recommendations.

Supported Platforms

Works with any AI assistant that supports the Agent Skills specification:

• Claude Code
• Cursor
• Windsurf
• Codex CLI
• Continue
• Gemini CLI
• And more

Project Structure

prism/
├── SKILL.md                              # Main skill instructions (both modules)
├── package.json                          # npm metadata
├── LICENSE.txt                           # MIT license
├── references/                           # Detailed generation guides
│   ├── PERSONA-GUIDE.md
│   ├── JOURNEY-MAP-GUIDE.md
│   ├── SYNTHESIS-GUIDE.md
│   ├── TEST-PLAN-GUIDE.md
│   ├── PHILOSOPHY-GUIDE.md
│   ├── TOKEN-GUIDE.md
│   ├── COMPONENT-GUIDE.md
│   ├── LAYOUT-GUIDE.md
│   ├── INTERACTION-GUIDE.md
│   ├── CANVAS-GUIDE.md
│   ├── FINSERV-LAYER.md
│   ├── FINSERV-UI-LAYER.md
│   └── QUALITY-CHECKLIST.md
├── scripts/
│   └── validate-output.js                # Output validator (zero dependencies)
└── assets/
    └── research-brief-template.json      # Structured input schema

Contributing

Contributions are welcome. Good places to start:

• Add new deliverable templates or reference guides
• Extend the financial services layer (or add a new domain layer)
• Add eval scenarios to the benchmark
• Report gaps where output quality could improve

Open an issue or a pull request — they're acknowledged quickly.

License

MIT