DYC

Eleven engineering workflow skills for AI agents — planning, review, debugging, UI, writing, research, and more. Sourced from tw93/Waza, adapted for multi-vendor runtimes.

中文

Features

Multi-vendor: one plugin.json source of truth, adapted to Claude Code, Codex, Qoder, and CodeBuddy via extensions + init.ts
11 skills covering the full engineering workflow: plan → build → review → debug → document
Writing rules for Chinese and English prose
9 MCP servers for browser automation, documentation, component libraries, and repository knowledge

Installation

# Claude Code
/plugin marketplace add mymx2/m2-plugins
/plugin install dyc@m2-plugins

# Codex
codex plugin marketplace add mymx2/m2-plugins
codex plugin add dyc@m2-plugins

# Qoder
vp run qoder:dyc

# CodeBuddy
/plugin marketplace add mymx2/m2-plugins
/plugin install dyc@m2-plugins

Prerequisites

Node.js ≥ 18
MCP servers auto-configured via mcp.json (playwright, chrome-devtools, deepwiki, context7, github, mdn, shadcn-vue, ai-elements-vue, tdesign-mcp-server)

Included Skills

Skill	Description
`think`	Turns rough ideas into approved, decision-complete plans before coding
`ui`	Produces distinctive, production-grade UI for pages, components, and typography
`check`	Reviews code diffs, PRs, release readiness, project audits, and doc proofreading
`hunt`	Finds root cause before applying fixes for errors, crashes, and regressions
`write`	Rewrites and polishes prose in Chinese or English, removes AI-like wording
`read`	Reads URLs and PDFs — fetches, summarizes, or converts to clean Markdown
`learn`	Six-phase research workflow turning unfamiliar material into publish-ready output
`health`	Audits agent instructions, config drift, hooks/MCP, and AI maintainability
`forge`	Plugin & skill lifecycle: create, adapt multi-vendor, find, install, author, validate
`chrome`	Drives a real Chrome browser via chrome-devtools-mcp for automation and analysis
`repowiki`	Generates a DeepWiki-style repository analysis report with Mermaid diagrams

Rules

File	Purpose
`rules/writing.md`	English coaching and Chinese anti-AI writing rules

MCP Configuration

Server	Description	Required Env Vars
`playwright`	Browser automation via Playwright MCP	None
`chrome-devtools`	Live Chrome browser control via chrome-devtools-mcp	None
`deepwiki`	Repository knowledge retrieval via DeepWiki	None
`context7`	Real-time library documentation via Context7	`CONTEXT7_API_KEY`
`github`	GitHub repository access via GitHub MCP Server	`GITHUB_PERSONAL_ACCESS_TOKEN`
`mdn`	MDN Web Docs retrieval via Mozilla MCP	None
`shadcn-vue`	shadcn-vue component library MCP	None
`ai-elements-vue`	AI Elements Vue registry MCP	None
`tdesign-mcp-server`	TDesign component library MCP	None

Environment Variables

Two MCP servers require API keys. Set them before starting your agent:

# Bash / Zsh (Linux / macOS)
export CONTEXT7_API_KEY="your-context7-api-key"
export GITHUB_PERSONAL_ACCESS_TOKEN="your-github-pat"

# PowerShell (Windows)
$env:CONTEXT7_API_KEY = "your-context7-api-key"
$env:GITHUB_PERSONAL_ACCESS_TOKEN = "your-github-pat"

:: CMD (Windows)
set CONTEXT7_API_KEY=your-context7-api-key
set GITHUB_PERSONAL_ACCESS_TOKEN=your-github-pat

To persist across sessions, add them to your shell profile (~/.bashrc, ~/.zshrc) or Windows system environment variables.

Variable	Where to Get It
`CONTEXT7_API_KEY`	Free at context7.com/dashboard
`GITHUB_PERSONAL_ACCESS_TOKEN`	GitHub → Settings → Developer settings → Personal access tokens

License

MIT