UI UX Pro Max
github.com/nextlevelbuilder/ui-ux-pro-max-skill- Category
- AI Agents
- Rank
- No. 1941Tools index
Previous survey · No. 1956 ·
- Pricing
- Open Source
- Type
- TOOL
- Use case
- Design & Media
- Interfaces
- Agent Skill / Plugin
- Builder
- nextlevelbuilder
- GitHub
- 133.8k stars
- Latest release
- v2.15.0
- Date
About
An AI agent skill that supplies design intelligence for building professional UI/UX across multiple platforms and frameworks. The repository ships the skill alongside a CLI, a gallery, and preview and project directories, with documentation in English, Vietnamese, and Chinese.
What it does
UI UX Pro Max gives a coding agent a local, searchable reference for interface decisions. Ask it for a spa landing page and it looks up a product category, candidate visual styles, a color palette, a page layout and a font pairing, then combines them with industry rules into one recommendation: layout sections, hex colors, fonts, effects, things to avoid and a pre-ship checklist. The lookup is plain keyword ranking over bundled spreadsheets run by a Python script, with no model call involved. A small Node installer copies it into Claude Code, Cursor, Copilot and many other assistants.
Why it's ranked here
The substance is in the data and the checks around it, not the marketing. A release gate validates every spreadsheet, runs unit tests, scores search relevance against separate calibration and held-out query sets, and smoke-tests each domain and stack before publishing. Reasoning rules carry provenance records with a confidence value and a review deadline. The output logic handles real conflicts, such as a dark style paired with a light palette. MIT licensed, with a Python search tool that needs nothing beyond the standard library.
What's good
The generator resolves light or dark mode once and makes the rest agree: a dark request swaps in a dark palette, derives dark surfaces while keeping brand colors, and drops advice that says to avoid dark mode. It computes WCAG contrast before picking a focus ring or outline-button text color, falling back when the brand color fails 4.5:1. Saved design systems go into a master file plus per-page overrides, and an existing master is never overwritten without an explicit force flag. Three optional dials for variance, motion and density adjust the output without rewriting the query.
Tradeoffs
The README calls the generator an AI-powered reasoning engine, but the code is keyword ranking plus lookup tables, so query wording decides the result and the skill itself tells the agent to retry with a narrower query when output is off-topic. Search returns three results by default. Stack advice is pinned to specific framework releases, so it is only as current as the maintainers keep it. The README also carries donation links and promotion for sibling projects, which is noise for anyone evaluating the tool.
How to use it well
Run the installer for your assistant with the dry-run flag first to see what it writes. Use the design-system mode once at the start of a new product to set palette, type and layout, persist it into the project, then use single-domain searches for targeted fixes such as focus states or badge wrapping. Treat results as a starting point to check against your brand. It does not build components or audit a running page; pair it with a browser test tool for that.
Technical notes+
cli/src/index.ts defines the uipro binary (commander) with init, versions, update and uninstall; init takes --ai, --global, --force, --dry-run and a GitHub --token for rate limits. cli/assets/scripts/core.py implements BM25 over CSV tables in a bundled data directory (style, color, chart, landing, product, UX, typography, icon, motion and Google Fonts domains plus per-stack tables), with MAX_RESULTS = 3, a synonym table, a short stopword list, per-domain score floors and a separate stack threshold, and a STACK_CURRENT_APPLICABILITY map pinning guidance to framework releases. cli/assets/scripts/design_system.py runs product, style, color, landing and typography searches, applies rules from the reasoning CSV via a reasoning contract module, resolves color mode with _resolve_color_mode, derives dark surfaces in _derive_dark_palette, filters dark-mode anti-patterns, and uses _contrast_ratio thresholds of 3 for the ring and 4.5 for outline text. cli/assets/data/data-provenance.json records status, confidence, sources and an SLA per entity. cli/package.json chains validate, unittest, relevance evaluation and smoke scripts into verify:data, run from prepublishOnly and from .github/workflows/tests.yml alongside pytest, Playwright and a gallery build.
Observed
- License
- MIT (LICENSE, cli/package.json, skill.json)
- Languages
- Python search and generator scripts; TypeScript installer CLI
- Install surface
- npm package ui-ux-pro-max-cli exposing the uipro command
- Interfaces
- Agent skill instructions plus a Python command-line search tool and a Node installer CLI
- Runtime dependencies
- Search tool requires Python 3 with no external packages, per the skill instructions
- Supported assistants
- skill.json lists Claude, Cursor, Windsurf, Copilot, Kiro, Codex, Gemini, OpenCode and others
- Search method
- BM25 keyword ranking over bundled CSV data, no model call (cli/assets/scripts/core.py)
- CI
- GitHub Actions runs a data verification gate, pytest, Playwright smoke tests and a gallery build (.github/workflows/tests.yml)
Read from README.md, cli/package.json, skill.json, .claude/skills/ui-ux-pro-max/SKILL.md, cli/src/index.ts, cli/assets/scripts/core.py, cli/assets/scripts/design_system.py, cli/assets/data/data-provenance.json, .github/workflows/tests.yml, LICENSE.
Tags
Media
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.