qwenkit
v1.8.0
Published
Qwen Code engineering kit — commands, skills, hooks, agents for professional software development
Maintainers
Readme
QwenKit Engineer
A comprehensive boilerplate for building professional software projects with Qwen Code — AI CLI coding agents.
What Is This
QwenKit Engineer is a production-ready configuration pack for Qwen Code that turns an AI coding agent into a full software engineering team.
Quick Start
Option 1: Install via npm (recommended)
npm install qwenkitThis creates .qwen/ in your project automatically via postinstall.
Option 2: Install CLI globally
npm install -g qwenkit
qk initUse qk init in any project to create .qwen/ with all commands, skills, hooks, and agents.
Migrate to another agent
Already have .qwen/ set up? Migrate your customized workflow to another AI agent:
qk migrate --claude # .qwen/ → .claude/
qk migrate --gemini # .qwen/ → .gemini/ (applies Gemini CLI transforms)
qk migrate --cursor # .qwen/ → .cursor/
qk migrate --from .claude --gemini # .claude/ → .gemini/- Skips files that already exist in target
- Applies per-platform transforms (tool remapping, hook event remapping)
- Updates
.qk.jsonagentDir and.gitignore - Copies
AGENTS.md→<AGENT>.mdat project root
Option 3: Manual setup
git clone https://github.com/qwenkit/qwenkit-engineer.git
cd your-project
node path/to/qwenkit-engineer/bin/qk.js initConfigure (optional)
Edit .qwen/.qk.json for your preferences:
{
"codingLevel": -1,
"statusline": "minimal",
"privacyBlock": true,
"locale": {
"thinkingLanguage": null,
"responseLanguage": null
}
}| Setting | Values | Description |
|---------|--------|-------------|
| codingLevel | -1 (auto), 0-3 | Output detail level by experience |
| statusline | full, compact, minimal, none | Status bar verbosity |
| privacyBlock | true, false | Protect sensitive files (.env, credentials) |
| locale.thinkingLanguage | null, "vi", "zh", ... | Agent reasoning language |
| locale.responseLanguage | null, "vi", "en", ... | Output language |
3. Install Skill Dependencies
cd .qwen/skills
# Linux/macOS
./install.sh
# Windows (PowerShell as Administrator)
.\install.ps1This installs FFmpeg, ImageMagick, Node.js packages, and Python packages needed by skills.
Commands
| Command | Purpose |
|---------|---------|
| /cook "task" | ⚡ Implement features step by step |
| /advanced-plan "scope" | ☠️ Deep plan for complex/risky work |
| /execute-plan [path] | 🚦 Execute approved plan task-by-task |
| /analyze [scope] | 🔍 Deep codebase understanding |
| /deep-research "question" | 🧭 Evidence-driven research |
| /architecture [scope] | 🏛️ Review boundaries and ownership |
| /refactor [scope] | ♻️ Safe structure improvements |
Pilot note:
refactornow also exists as a skill atskills/refactor/, while legacy slash command files remain for compatibility during migration. |/split-file [path]| 📁 Split large files safely | |/optimize [hotspot]| ⚡ Performance improvements | |/advanced-debug "symptom"| 🧩 Hard bug investigation | |/advanced-enhance "feature"| 🚀 Complex feature changes | |/security [scope]| 🔐 Security audit and hardening | |/review:codebase| 👁️ Codebase scan and review | |/test| Write and run tests | |/plan| Create implementation plan | |/bootstrap| Initialize new project | |/preview [path]| View files or generate diagrams | |/coding-level| Set output detail level | |/journal| Write development journal | |/kanban| View task board | |/watzup| Check status | |/worktree| Git worktree management | |/use-mcp| MCP server integration | |/ask| Quick questions | |/qk-help [topic]| QwenKit usage guide |
Full reference: commands/README.md
Quick pick: commands/CHEATSHEET.md
Detailed guides: commands/REFERENCES/
Skills (50+)
Core Development
| Skill | Purpose |
|-------|---------|
| backend-development | Node.js, Python, Go APIs, databases, microservices |
| frontend-development | React, Next.js, Vue, TypeScript, performance |
| mobile-development | React Native, Flutter, Swift, Kotlin |
| databases | PostgreSQL, MongoDB, schema design, migrations |
| web-frameworks | Next.js, SSR, caching, monorepos |
| ui-styling | shadcn/ui, Tailwind CSS, themes |
Quality & Safety
| Skill | Purpose |
|-------|---------|
| code-review | Review code for correctness, security, quality |
| debug | Root cause analysis, systematic debugging |
| fix | Bug fixing with workflow patterns |
| web-testing | Playwright, Vitest, E2E, load testing |
| security | OWASP, vulnerability scanning |
AI & Automation
| Skill | Purpose |
|-------|---------|
| ai-artist | Generate images via AI (Nano Banana) |
| ai-multimodal | Analyze images/audio/video with Gemini |
| chrome-devtools | Browser automation, performance analysis |
| threejs | 3D web apps with Three.js (556 examples) |
| shader | GLSL fragment shaders for procedural graphics |
DevOps & Infrastructure
| Skill | Purpose |
|-------|---------|
| devops | Cloudflare, Docker, GCP, Kubernetes, CI/CD |
| git | Git operations, conventional commits, PRs |
| mcp-management | MCP server discovery, tools, integration |
| mcp-builder | Build MCP servers for LLM integration |
| better-auth | Authentication with Better Auth |
Design & UX
| Skill | Purpose |
|-------|---------|
| ui-ux-pro-max | UI/UX design intelligence, 50 styles, 21 palettes |
| web-design-guidelines | Review UI code for best practices |
| frontend-design | Polished interfaces from designs |
Productivity
| Skill | Purpose |
|-------|---------|
| cook | Task execution with internal planning |
| planning | Technical planning and architecture |
| plans-kanban | Plan dashboard and tracking |
| research | Comprehensive technical research |
| docs-seeker | Search library documentation via context7 |
| repomix | Pack repositories for AI analysis |
| scout | Fast codebase exploration |
| brainstorm | Solution brainstorm with trade-offs |
| copywriting | Conversion copywriting, headlines, CTAs |
| mintlify | Build documentation sites with Mintlify |
| mermaidjs-v11 | Create diagrams (flowcharts, sequence, ER) |
Specialized
| Skill | Purpose |
|-------|---------|
| shopify | Build Shopify apps, extensions, themes |
| payment-integration | SePay, Stripe, Paddle, subscriptions |
| media-processing | FFmpeg, ImageMagick, background removal |
| document-skills | DOCX, PPTX, PDF processing |
| markdown-novel-viewer | Distraction-free markdown reading |
| skill-creator | Create or update Qwen Code skills |
Full list: skills/README.md
Agents
| Agent | Purpose |
|-------|---------|
| planner | Research, analyze, create implementation plans |
| fullstack-developer | Execute implementation phases |
| code-reviewer | Code review with edge case detection |
| debugger | Investigate issues, diagnose problems |
| tester | Run tests, analyze coverage, validate builds |
| researcher | Comprehensive technical research |
| docs-manager | Technical documentation management |
| ui-ux-designer | UI/UX design and wireframes |
Hooks
Hooks run automatically during Qwen Code sessions.
| Hook | Trigger | Purpose |
|------|---------|---------|
| session-init | SessionStart | Project detection, env setup |
| subagent-init | SubagentStart | Inject context to subagents |
| privacy-block | PreToolUse | Block access to sensitive files |
| scout-block | PreToolUse | Block heavy directory scans |
| dev-rules-reminder | UserPromptSubmit | Remind development rules |
| descriptive-name | PreToolUse | Ensure descriptive script names |
| usage-context-awareness | UserPromptSubmit/PostToolUse | Usage limit tracking |
| cook-after-plan-reminder | PostToolUse | Remind to cook after planning |
| post-edit-simplify-reminder | PostToolUse | Remind to simplify after edits |
Toggle hooks in .qwen/.qk.json:
{
"hooks": {
"session-init": true,
"privacy-block": true,
"scout-block": false
}
}Output Styles
Set your coding experience level in .qwen/.qk.json:
{ "codingLevel": -1 }| Level | Value | Style | |-------|-------|-------| | Auto-detect | -1 | Qwen detects from your usage | | Beginner | 0 | Verbose explanations, hand-holding | | Intermediate | 1 | Moderate detail, some assumptions | | Senior | 2 | Concise, expects experience | | Lead | 3 | Minimal, high-signal | | God-mode | 4-5 | Code only, zero explanation |
Output style files: output-styles/
Config Reference
Full config in .qwen/.qk.json:
{
"$schema": "./schemas/qk-config.schema.json",
"codingLevel": -1,
"statusline": "minimal",
"privacyBlock": true,
"docs": { "maxLoc": 800 },
"plan": {
"namingFormat": "{date}-{issue}-{slug}",
"dateFormat": "YYMMDD-HHmm",
"issuePrefix": "GH-"
},
"paths": {
"docs": "docs",
"plans": "plans"
},
"locale": {
"thinkingLanguage": null,
"responseLanguage": null
},
"hooks": {},
"project": {
"type": "auto",
"packageManager": "auto",
"framework": "auto"
},
"gemini": { "model": "gemini-3-flash-preview" },
"skills": { "research": { "useGemini": true } }
}Schema: schemas/qk-config.schema.json
Ignore Patterns
.qwen/.qkignore blocks heavy directories from scout operations:
# Default blocked patterns
node_modules
.git
.venv
__pycache__
dist
buildProject Structure
your-project/
├── .qwen/
│ ├── agents/ # Agent definitions (8 files)
│ ├── commands/ # Slash commands (25 + REFERENCES/)
│ │ ├── REFERENCES/ # Detailed checklists (11 files)
│ │ ├── README.md # Full command reference
│ │ └── CHEATSHEET.md # Quick selection guide
│ ├── hooks/ # Session hooks (9 hooks)
│ │ ├── lib/ # Hook utilities
│ │ └── notifications/ # Discord/Slack/Telegram alerts
│ ├── output-styles/ # Coding level configs (6 levels)
│ ├── rules/ # Development rules
│ ├── schemas/ # Config JSON schema
│ ├── scripts/ # Core utility scripts
│ ├── skills/ # 50+ capability skills
│ ├── .qk.json # Project configuration
│ ├── .qkignore # Scout ignore patterns
│ ├── .env.example # Environment template
│ ├── settings.json # Qwen Code settings
│ └── statusline.* # Status line scripts
└── ... your codeCommon Workflows
Build New Feature
/cook "add user authentication"
→ /test
→ /review:codebase
→ git commit + pushFix Bug
/advanced-debug "users get 500 error on login"
→ /cook "fix root cause"
→ /test
→ /review:codebaseUnderstand Codebase
/analyze
→ /cook "implement X" (with full context)Complex Feature
/advanced-plan "build payment system"
→ Review plan
→ /execute-plan
→ /test → /security → /review:codebasePerformance
/analyze [hotspot]
→ /optimize "slow API endpoint"
→ /testLicense
MIT — free for personal and commercial use.
See LICENSE for full terms.
Contributing
- Fork the repository
- Create a feature branch
- Make your changes
- Test thoroughly
- Submit a pull request
Acknowledgments
- Built for Qwen Code — AI CLI coding agents
- Skills and commands designed for professional software engineering workflows
- Inspired by best practices in modern development workflows
