gate

AI Skill Routing System

When a user makes a request, route to the corresponding skill according to the following rules.

Routing Decision Rules

Rule 1: Identify user intent

Analyze the user input and identify the following intent types:

Intent type	Keyword examples	Routing target
Site building	build, create new, from scratch	`/gate-discuss`
Frontend development	page, component, style, UI	`/gate-frontend`
SEO	search engine, SEO, optimization	`/gate-seo`
UI design	design, interface, prototype, UX	`/ui-ux-pro-max`
Component design	component, reuse, component library	`/component-design`
Styling	Tailwind, CSS, style	`/styling-tailwind`
Code review	review, review, check code	`/code-review`
Debugging	bug, debug, issue, error	`/diagnosing-bugs`
Testing	test, TDD, unit test	`/tdd`
Refactoring	refactor, optimize code, code structure	`/ai-refactoring`
Product	PRD, requirements, user story	`/prd-development`
Optimization	optimize, performance, speed	`/gate-optimize`

Rule 2: Check project status

Before routing, check the current project status:

# 检查是否存在项目
ls package.json 2>/dev/null    # Node.js 项目
ls pyproject.toml 2>/dev/null  # Python 项目
ls Cargo.toml 2>/dev/null      # Rust 项目

# 检查是否有 Git 仓库
git status 2>/dev/null

# 检查是否有部署配置
ls vercel.json 2>/dev/null     # Vercel 部署
ls netlify.toml 2>/dev/null    # Netlify 部署

Status determination：

No project → new project flow (/gate-discuss）
Has project but not completed → continue development flow
Project already launched → optimization flow (/gate-optimize）

Rule 3: Routing priority

When multiple skills match, choose according to the following priority:

Skill explicitly specified by the user → call directly
Skill most relevant to the current stage → based on project status + next field selection
Skill that best matches user intent → based on keyword matching
Default routing → ask the user for specific requirements

Rule 4: Recommend next step

After calling a skill, read its next field and automatically recommend it to the user.

The complete routing information for each skill (trigger / next / dependencies) is in the frontmatter of its own SKILL.md. There is no need to maintain a copy here.

Routing decision flow

用户输入
    ↓
分析意图（关键词匹配）
    ↓
检查项目状态
    ↓
选择最匹配的技能
    ↓
调用技能
    ↓
执行任务
    ↓
更新项目状态
    ↓
推荐下一个技能

Usage examples

Example 1: A new user wants to build a site

用户：我想搭建一个个人网站
AI：分析意图 → 建站
AI：检查项目状态 → 无项目
AI：路由到 → /gate-discuss

Example 2: The user has a project that needs optimization

用户：我的网站加载太慢了
AI：分析意图 → 性能优化
AI：检查项目状态 → 项目已部署
AI：路由到 → /gate-perf

Example 3: The user needs design help

用户：我想设计一个登录页面
AI：分析意图 → UI 设计
AI：检查项目状态 → 有项目
AI：路由到 → /ui-ux-pro-max

Notes

Prefer using gate- skills*: The site building flow and optimization flow should preferentially use the corresponding gate skills
Call skills directly: Design, engineering, AI, and product skills can be called directly
Check dependencies: Before calling a skill, check whether there are dependency skills
Update status: After completing a skill, update the project status
Recommend next step: Recommend the next related skill after completion