Forkcast AI Skills #5: ui-ux-pro-max-skill, ruflo, cangjie-skill, Vibe-Skills, agent-rules-books

Episode Summary#

This episode covers 30 AI-powered tools and skills including a complete UI/UX design framework with 1000+ templates, an AI video generation engine, and skills for turning books into agent capabilities. Features MCP servers, code analysis toolkits, and production-ready agent frameworks.

Repositories Covered#

ui-ux-pro-max-skill#

AI coding agents generate poor UI designs because they lack structured design intelligence — color palettes, component patterns, and anti-pattern avoidance are not in their training data Covers React, HTML5, Tailwind CSS, mobile UI. Supports Trae, Copilot, Codex, Cursor, Windsurf, Claude Code, Kiro, Qoder, Antigravity. v2.13.0 with 210 commits. 47 tags, 17 branches. Generates CSS variables, palette modes, and anti-pattern filters. MIT license.

ruflo#

Building multi-agent AI systems requires stitching together memory, RAG, workflow coordination, and harness interfaces from scratch for each project 7,302 commits, 490 branches, 1,612 tags. Adaptive memory, self-learning intelligence, RAG integration. Native support for Claude Code, Codex, Hermes. 8K forks. MIT license. TypeScript/npm ecosystem.

cangjie-skill#

Valuable methodologies and frameworks in books, courses, and podcasts are locked in passive formats — people read them once and forget the actionable patterns Seven-stage pipeline: extraction, chunking, triple verification, DIGEST.md long-form output. Supports books, videos, podcasts, courses with field mapping. AGPL-3.0. 42 commits, 829 forks. Bilingual (Chinese/English) with community WeChat group.

Vibe-Skills#

AI agents with multiple skills have no built-in intelligence for deciding which skill to activate for a given task — manual selection defeats the purpose of autonomous agents 1,400 commits, 85 branches, 17 tags. 10 contributors. Bundled skills library, adapter system, agent templates. Python 62.5%, PowerShell 30%. 14 releases. Supports Codex, Claude Code ecosystem.

agent-rules-books#

AI coding agents produce code that violates fundamental software engineering principles because they have no structured rules from established methodology books Covers 14 books: Clean Code, Refactoring, DDD (distilled + full + implementing), Clean Architecture, Code Complete, DDIA, Patterns of Enterprise Application Architecture, Release It, A Philosophy of Software Design, Refactoring Guru. 52 commits. MIT license.

docmd#

Documentation tools are either bloated React frameworks that take minutes to build, or static generators that lack search, AI, and version control integration 2,423 commits, 83 tags. i18n support, API docs, Docker, semantic search, AI plugin, OpenAPI support, PWA, version control. Node.js/TypeScript. MIT license. 119 forks.

skills#

Business and marketing knowledge from bestselling books is inaccessible to AI agents — agents can code but cannot apply proven business frameworks like Lean Startup or Blue Ocean Strategy 50 individual skills + 12 metaskills (guided journeys). Covers business, marketing, UX, code quality from books. Works with Claude Code, Codex, Cursor. v1.5.1 with 7 releases. 69 commits. agentskills.io compatible.

one-skill-to-rule-them-all#

Skills degrade over time because they cannot learn from user corrections — every mistake or preference adjustment must be manually coded into the skill file v2.0.0 with progressive disclosure and self-improvements from six months of continuous use. Pre-Flight Principle: embedded commands must be executed once against real data before saving. 38 commits, 150 forks. CC BY 4.0. Domain-agnostic.

n8n-as-code#

Building n8n workflows through the UI is slow and error-prone for agents — there’s no programmatic way to generate, version-control, and deploy workflow automations 537 nodes with full schemas, 7,700+ templates. Git-like sync, TypeScript workflows. VS Code extension with AI assistant. 1,430 commits, 67 branches, 361 tags. MIT license. 173 forks.

skillpack#

Deploying AI agent skills to a team requires manual setup per machine — there’s no standardized packaging format that works across environments 197 commits, 15 branches. Skillpack creator skill, web UI, templates, tests. HTTP request timeout increased to 10 minutes. TypeScript runtime. MIT license. 106 forks. Discord community.

SkillNet#

There is no systematic way to evaluate whether an AI skill actually works well or to discover connections between related skills across different agents Knowledge-graph-based skill network. 175 commits, 130 forks. Automated skill review for SKILL.md pull requests via GitHub Action. MIT license. OpenKG integration. Academic research project from Zhejiang University.

ai-factory#

Setting up the right AI context, prompts, and spec-driven workflows for each project takes hours of configuration before writing any actual code 513 commits, 38 tags. v2.x branch. Spec-driven development approach. 92 forks. MIT license. Active development with budget-tracking features for AI agent loops.

app-store-connect-cli-skills#

App Store Connect operations like TestFlight management, screenshot uploads, and release scheduling require repetitive manual work through Apple’s web interface 132 commits, 51 forks. Packaged for Codex and Claude Code. Supports TestFlight builds, metadata, screenshots, phased releases. v3.5.0. MIT license. Works with asc CLI.

aiden#

Desktop automation for AI agents lacks a unified platform — browser control, terminal access, and workflow management each require separate tools with no shared memory 1,598 commits, 37 branches, 73 tags. Browser control, terminal execution, persistent memory, recovery systems. TypeScript 93.9%. AGPL-3.0. Built solo. Cross-platform with Windows CI.

compass-skills#

AI agents lack a personal task management layer that aligns their capabilities with individual user workflows and preferences over time 54 commits, 57 forks. Task clarifier, run-history tracking, academic style analysis. MIT license. Local-first design. Chinese/English documentation.

flutter-ai-rules#

AI coding agents consistently produce non-idiomatic Flutter code because they lack Flutter-specific design patterns, widget lifecycle rules, and Dart conventions Works with Claude Code, Codex CLI, Cursor, Windsurf, Antigravity. Dart/Flutter specific. Covers widget patterns, state management, testing. 2 branches, MIT license.

claude-skill-registry#

Discovering and installing the right Claude Code skill requires manually searching GitHub with no central index or quality filtering Web-based registry at majiayu000.github.io. HTML/JS frontend. Auto-generated from core artifacts. GitHub Pages deployment. skill-registry topic.

ai-toolkit#

Developers using multiple AI coding platforms need to maintain separate skill sets for each — Claude, Cursor, Windsurf, Copilot, Gemini, and others each have different formats 94 skills, 44 agents. Multi-platform: Claude, Cursor, Windsurf, Copilot, Gemini, Cline, Roo Code, Aider, Augment, Antigravity, Codex CLI, opencode. Python-based. MCP support.

akm#

Managing the growing collection of AI agent knowledge files, memories, and skills across projects becomes unwieldy without a dedicated CLI tool TypeScript CLI. MCP support, opencode integration. Agent kit manager. Package manager for AI agent assets. 52 stars, actively maintained.

European-Parliament-MCP-Server#

Accessing European Parliament data for political analysis requires navigating complex official databases with no programmatic API for AI agents TypeScript MCP server. Covers parliamentary debates, votes, legislation. OSINT-capable. 22 stars. EU geopolitical analysis. Political science research tool.

mobile-ai-agents#

Mobile developers lack AI-specific skills for Android, iOS, Flutter, React Native, Unity, and Unreal — generic coding agents produce poor mobile code Covers Android, iOS, Flutter, React Native, Unity, Unreal. AI agents, prompts, workflows. JavaScript-based. Code review, bug fixing, performance optimization skills.

skills-tracker#

The AI skills ecosystem is growing so fast that developers cannot keep up with new repos — there’s no automated way to monitor the trend Auto-updated Python tracker. Monitors new GitHub skills repos. Machine learning, developer tools, awesome-list, agent-skills topics. Live data pipeline.

claude-skill-registry-core#

The Claude skill ecosystem has grown so large that finding high-quality skills requires a deduplicated, searchable index with automated curation Python-based search index. GitHub Pages deployment. Automation for skill discovery. 23 stars. Companion to the main registry. Deduplication engine.

euparliamentmonitor#

European Parliament proceedings span 24 official languages and hundreds of committees, making comprehensive political monitoring impossible for individuals AI-generated political news in 14 languages. PHP backend. Democratic transparency platform. Autonomous agentic newsroom. Open data powered.

riksdagsmonitor#

Swedish parliamentary proceedings are scattered across official databases with no aggregated intelligence layer for citizens and researchers 9 stars, TypeScript/HTML. AI-generated political news. Democratic transparency. Evidence-based analysis. Autonomous agentic newsroom. Open data integration.

skill-vault#

Finding and installing the right Claude Code skill requires manually browsing GitHub — there’s no automated discovery and installation tool HTML/JS skill discovery tool. Automated search, GitHub search integration, security scanner. 2 stars. Meta-skill for skill management. Claude Code compatible.

career-copilot#

Job hunting involves repetitive manual work across job boards, resume tailoring, and application tracking that consumes hours daily Docker support, CLI, MCP server. Next.js frontend. Langfuse + LangGraph. Google ADK compatible. Resume generator. Interview prep.

context-kit#

Developers lack a centralized kit of productivity plugins and skills that work together across Copilot, Cursor, and other AI tools Python-based. Optimization, Copilot integration, AI agents, AI tools, context engineering. GitHub Copilot compatible.

apple-dev-skills#

Apple platform development with Swift, SwiftUI, and Xcode requires domain-specific knowledge that generic AI agents lack — TestFlight, App Store Connect, and SwiftData patterns are not well represented TestFlight, Swift, iOS, Apple, Xcode, MCP, Cursor, SwiftUI, App Store Connect, SwiftData. Multi-agent support. JavaScript/TypeScript.

appsec-skills#

Security analysis requires specialized knowledge across SAST, threat modeling, and serialization audits that generic coding agents lack Covers penetration testing, ethical hacking, risk management. Built for Gemini CLI and other agents. Evidence-first security workflows. JavaScript based.

Watch#

• Video: https://www.youtube.com/watch?v=PENDING_UPLOAD
• Cover: https://i.ytimg.com/vi/PENDING_UPLOAD/hqdefault.jpg

Notes#

Transcript and notes will be added from Forkcast output artifacts.