Analyzthis_Design

7 AI design personas for Cursor. Run structured UX critiques and multi-phase ideation sessions — right inside your AI chat.

Install

Works with Cursor, Claude Code, and Codex CLI.

# Cursor (default)
npx analyzthis_design

# Claude Code
npx analyzthis_design --target claude

# Codex CLI
npx analyzthis_design --target codex

# All three at once
npx analyzthis_design --target all

Skills are copied to the correct directory for each tool and available immediately.

| Tool | Install path | |---|---| | Cursor | ~/.cursor/skills/ | | Claude Code | ~/.claude/commands/ | | Codex CLI | ~/.codex/skills/ |

What gets installed

2 Orchestrating Skills

| Skill | What it does | |---|---| | /design-critic | 4-persona critique → returns a SHIP / REVISE / BLOCK verdict with a Composite Score | | /ux-ideator | 6-phase ideation workflow → produces two competing IA concepts, deliberates between them, and delivers an implementation-ready wireframe |

7 Individual Persona Skills

You can also invoke each persona directly.

| Skill | Persona | Role | |---|---|---| | /arjun | UX Agent | Scores designs on the UX Honeycomb (7 dimensions, A–F) | | /meera | Business Agent | Evaluates retention, ARR, GTM lever, adoption risk by segment | | /priya | Feasibility Agent | T-shirt sizes engineering effort using a 2-axis model, flags state machine traps | | /zara | Delight Agent | Picks exactly ONE peak delight moment (or says "speed is the craft") | | /noor | IA Architect | Produces Concept A — minimalist, progressive disclosure, ≤3 nav levels | | /anuj | Power-User Advocate | Produces Concept B — dense, bulk actions, keyboard shortcuts | | /raj | Arbitrator | Resolves stalemates using 5 ranked product principles — never speaks first |

1 Context Template

| Skill | Purpose | |---|---| | /design-personas | Session context template — fill in once before any session so all 7 personas have project-specific grounding |

How to use

Quick design critique

Open a Cursor chat and type:

/design-critic

[Paste your screen description, mockup link, or component spec]

Arjun, Meera, Priya, and Zara each evaluate the design independently. You get a Composite Score out of 20 and a clear SHIP / REVISE / BLOCK verdict.

Full UX ideation from scratch

/design-personas
[Fill in the session context template]

/ux-ideator
[Describe the feature you need to design]

The 6-phase workflow runs:

1. Meera reframes the request as a business outcome
2. Noor + Anuj + Arjun audit the existing IA for gaps
3. Noor produces Concept A (minimalist), Anuj produces Concept B (dense)
4. They deliberate — Raj arbitrates if they deadlock
5. Arjun validates the synthesized concept on the UX Honeycomb
6. Zara adds one delight moment (or says it's a working surface, skip)
7. Priya sizes the engineering effort and flags risks

Single persona

/arjun review this form flow — [description]
/priya estimate this feature — [description]
/zara find the delight moment in this onboarding — [description]

CLI commands

# Install for Cursor (default)
npx analyzthis_design

# Install for Claude Code
npx analyzthis_design --target claude

# Install for Codex CLI
npx analyzthis_design --target codex

# Install for all tools at once
npx analyzthis_design --target all

# Force overwrite existing skills
npx analyzthis_design --target all --force

# Remove all installed skills (from a specific tool)
npx analyzthis_design remove --target cursor
npx analyzthis_design remove --target all

# Check what's installed
npx analyzthis_design list
npx analyzthis_design list --target all

How it works

This package copies 10 SKILL.md files into your ~/.cursor/skills/ directory. Cursor automatically makes them available as /skill-name commands in any chat session. The skills are plain markdown — no backend, no API keys, no runtime dependencies.

Each persona has:

• A defined lens (what they evaluate and how)
• A mandatory output format (so you get structured, comparable output)
• Canonical failure patterns (real anti-patterns they watch for)
• Failure modes (where the persona itself can go wrong — so the AI self-corrects)

Requirements

• Cursor with Agent Mode
• Node.js 16+

License

MIT