agent-park/CLAUDE.md

# CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

## Development Commands

### Build & Run
```bash
pnpm dev          # Start development server (Next.js 15)
pnpm build        # Build for production
pnpm start        # Start production server
pnpm lint         # Run ESLint
```

### Testing
```bash
pnpm test         # Run Vitest unit tests
pnpm test:e2e     # Run Playwright E2E tests
```

### Database
```bash
pnpm prisma migrate dev   # Run database migrations
pnpm prisma migrate dev --name init    # Create initial migration
pnpm prisma db seed       # Seed database (uses ts-node)
pnpm prisma studio        # Open Prisma Studio for database inspection
```

## Architecture Overview

This is a **Next.js 15 multilingual AI project navigation website** using the App Router architecture with the following key components:

### i18n Architecture (next-intl)
- **Locales**: `zh` (default) and `en`
- **Route pattern**: `/{locale}/path` (always prefixed with locale)
- **Middleware**: `src/middleware.ts` handles locale detection and routing
- **i18n config**: `src/i18n/request.ts` loads locale messages from `src/messages/{locale}.json`
- **Messages**: Translation files at `src/messages/zh.json` and `src/messages/en.json`

### App Router Structure
```
src/app/
├── [locale]/           # Locale-scoped routes
│   ├── page.tsx        # Home page
│   ├── projects/       # Projects listing and details
│   │   ├── page.tsx    # Projects list
│   │   └── [id]/       # Individual project details (slug-based)
│   └── layout.tsx      # Locale layout (header, footer)
├── api/                # API routes (no locale prefix)
│   └── webhook/projects/route.ts  # Webhook for project ingestion
└── layout.tsx          # Root layout
```

### Database (Prisma + PostgreSQL)
- **Provider**: Neon (serverless PostgreSQL)
- **Schema**: `prisma/schema.prisma` defines models: `Project`, `Tag`, `ExternalLink`, `ProjectTag`
- **Enums**: `ProjectStatus` (ACTIVE/ARCHIVED), `LinkType` (WEBSITE/GITHUB/HUGGINGFACE/PAPER)
- **Client singleton**: `src/lib/prisma.ts` exports Prisma client instance
- **Multilingual fields**: Most models have `name`/`nameEn`, `description`/`descriptionEn`, `content`/`contentEn` pairs
- **Cascade deletions**: `ExternalLink` and `ProjectTag` use `onDelete: Cascade` - deleting a project automatically cleans up its links and tag connections
- **Key constraints**:
  - `Project.slug`: Unique
  - `Tag.name`: Unique (tag names are globally unique)
  - `Tag.slug`: Unique
  - `ExternalLink`: `@@unique([projectId, url])` (each project can't have duplicate URLs)
- **Indexes**: `idx_project_status_createdAt`, `idx_project_slug`, `idx_tag_slug`, `idx_link_projectId`, `idx_link_type`, `idx_link_type_url` (composite for efficient URL-based deduplication)

#### Neon Database Setup
- **Dashboard**: https://console.neon.tech
- **Connection String Format**: `postgres://[user]:[password]@[host]/[database]?sslmode=require`
- **Vercel Integration**: Set `DATABASE_URL` environment variable in Vercel Dashboard (do NOT use `vercel.json` env references)
- **Free Tier**: 0.5GB storage, 300 hours compute/month
- **Run Migrations**: After deployment, run `pnpm prisma db push` or use Neon's SQL Editor to create tables

### Webhook Deduplication Strategy
The webhook at `src/app/api/webhook/projects/route.ts` implements a **multi-level deduplication** strategy to prevent duplicate projects:
1. **GitHub URL exact match** (highest priority) - via `ExternalLink` table using `idx_link_type_url` index
2. **Website URL exact match** - via `ExternalLink` table using `idx_link_type_url` index
3. **slug match** (fallback) - via `Project.slug` field

When updating an existing project, the webhook:
- Updates all project fields (name, description, content, status, source)
- Replaces all tags (deletes old `ProjectTag` connections via `ProjectTag` table, creates new ones)
- Replaces all links (deletes old `ExternalLink` entries, creates new ones)

**Tag handling special case**: Due to `Tag.name` unique constraint, tag upsert follows:
1. First try to find existing tag by name
2. If not found, try upsert by slug
3. If slug conflicts, use the existing tag with that slug

### Data Fetching (Server-Side)
- **Location**: `src/hooks/useProjects.ts` (server functions, not React hooks)
- **Functions**: `getProjects()`, `getProjectBySlug()`, `getAllTags()`, `getTagsWithProjectCounts()`
- **Usage**: Directly called in Server Components and route handlers
- **Query transformation**: `getProjects()` and `getProjectBySlug()` flatten the `ProjectTag` junction table structure to return tags directly
- **ISR**: Project detail pages use `export const revalidate = 300` (5 minutes) at `src/app/[locale]/projects/[id]/page.tsx`

### Validation (Zod)
- **Schemas**: `src/lib/validations.ts` defines all Zod schemas
- `ProjectInputSchema`: Validates incoming project data (1-10 tags, 1-10 links required)
- `WebhookPayloadSchema`: Validates webhook requests with API key (1-100 projects per request)
- `ProjectQuerySchema`: Validates query parameters (search, tags, status, page, limit)

### Styling (Tailwind CSS)
- **Neo-brutalism design**: Sharp corners (0px radius), bold borders (4px shadows), hard edges
- **Theme colors**:
  - Primary: Gold (#FFD700)
  - Secondary: Orange (#ff6f00)
  - Background light: #F5F2EB, dark: #121212
  - Surface light: #FFFFFF, dark: #1E1E1E
- **Typography**: Space Mono (headings), Inter (body)
- **Dark mode**: Class-based with `dark:` prefix
- **Config**: `tailwind.config.ts` extends theme with custom colors, shadows, and animations

### Content Rendering
- **Markdown**: Project content fields support Markdown via `react-markdown`
- **Plugins**: `rehype-raw`, `rehype-sanitize`, `rehype-shiki`, `remark-gfm`
- **Usage**: `ProjectDetail` component renders `content`/`contentEn` as Markdown

### UI Components
- **Radix UI primitives**: `@radix-ui/react-slot`, `@radix-ui/react-navigation-menu`, `@radix-ui/react-dropdown-menu`, `@radix-ui/react-separator`
- **Lucide React icons**: Used throughout the app (package imports optimized via `experimental.optimizePackageImports`)
- **Custom components**: `src/components/` organized by domain
  - `layout/`: Header, Footer, AnnouncementBar
  - `locale/`: LocaleSwitcher
  - `project/`: ProjectCard, ProjectList, ProjectDetail, ProjectSidebar, RelatedProjects, TagCloud, ExternalLinkCard, ShareButtons, MarkdownContent, GitHubBadges, GitHubTextStatsCard
  - `search/`: SearchBar
  - `ui/`: Base UI components (buttons, cards, etc.)

### Next.js Configuration
- **next.config.js**:
  - `next-intl` plugin wrapper for i18n
  - Image domains: localhost, *.anthropic.com, img.shields.io
  - Lucide-react package import optimization
- **tsconfig.json**: ES2017 target, strict mode enabled, noUncheckedIndexedAccess enabled
- **Testing**: Vitest for unit tests, Playwright for E2E tests (configured but not extensively used yet)

### Project Discovery System
项目发现系统是自动化探索和收录 AI 项目的核心功能，采用**双 Agent 协作架构**实现上下文隔离：

#### 架构组件
1. **自定义 Agents** (`.claude/agents/`):
   - `content-explorer-agent`: 项目内容探索专家，批量探索项目并生成结构化数据
     - 使用 `agent-browser` 子任务并行探索 GitHub 项目
     - 应用严格的内容质量标准（客观描述、避免营销术语、不写入动态数据）
     - 生成符合 `ProjectInputSchema` 的 JSON 数据
   - `api-submitter-agent`: API 提交专家，处理探索结果的提交和状态更新
     - 批量标记任务为 IN_PROGRESS
     - 提交探索数据到完成 API
     - 自动重试失败的提交（指数退避，最多3次）

2. **Claude Commands** (`.claude/commands/`):
   - `/discover-projects`: 主命令，协调探索和提交流程
     - 参数解析（任务数量、批次大小）
     - 分批处理（默认每批3个任务）
     - Agent 调度和进度显示
     - 结果汇总和错误报告

3. **API Endpoints** (`src/app/api/discovery/`):
   - `POST /api/discovery/tasks`: 创建新的探索任务（支持批量）
   - `GET /api/discovery/tasks`: 获取待处理任务列表（支持 status/limit/offset 过滤）
   - `PATCH /api/discovery/tasks/{id}`: 更新任务状态
   - `POST /api/discovery/tasks/{id}/complete`: 完成任务并提交项目数据
   - `GET /api/webhook/check-duplicates`: 检查项目是否已存在（URL 去重）

4. **Database Model**:
   - `ProjectDiscoveryTask`: 任务追踪表
     - 状态: PENDING → IN_PROGRESS → COMPLETED/FAILED
     - 原始数据: `sourceUrl`, `sourceType`
     - 探索结果: `explorationData` (JSON), `explorationSummary`
     - 错误处理: `errorMessage`, `retryCount`, `lastRetryAt`
     - 索引: `idx_task_status_created`, `idx_task_source_url`, `idx_task_project_id`

#### 数据流转
```
用户输入 URL → 创建 PENDING 任务 → /discover-projects 命令
    ↓
分批获取任务（每批3个）
    ↓
Content Explorer Agent (并行探索) → 探索结果 JSON
    ↓
API Submitter Agent (提交到生产环境 API)
    ↓
更新任务状态 → COMPLETED/FAILED
```

#### 质量标准
- **数据模板**: `.claude/schemas/project-content-template.md`
- **描述要求**: 清晰说明功能、突出价值、避免营销术语、10-500字
- **内容要求**: 从 README 提取并重新组织、不机械翻译、符合中文表达习惯
- **链接要求**: 必须包含 GITHUB 链接、所有链接可访问
- **标签要求**: 1-10 个标签、技术/应用/状态分类
- **动态数据处理**: Star/Fork 数量等动态数据不写入内容，使用 GitHub Badge 显示

#### 环境变量
- `WEBHOOK_API_KEY`: 生产环境 API 密钥（必需，用于认证）

#### 使用示例
```bash
# 处理默认10个任务（每批3个）
/discover-projects

# 处理指定数量的任务
/discover-projects 5

# 自定义批次大小
/discover-projects 9 --batch=2

# 处理所有待处理任务
/discover-projects all --batch=5
```

## MCP Servers Usage (按需使用)

1. **context7**: 不确定 API 用法时查阅最新文档
2. **chrome-devtools-mcp**: 查看页面效果、调试 UI 修复 BUG
3. **web-search-prime**: 默认联网搜索工具
4. **vision-mcp-server**: 图片/视频理解

## Git Commits

Git 提交信息遵循约定式提交格式（详见上方 Code Quality & Standards → Git Commit Conventions）。

## Development Workflow

- Before viewing the page, check if there's already a project running at localhost:3000. If yes, access it directly; if no, then run `pnpm dev`
- The dev server runs with hot-reload enabled for fast iteration
- The middleware handles locale detection and routing automatically - no manual locale configuration needed
- When adding new translations, update both `src/messages/zh.json` and `src/messages/en.json`
- Database changes require running `pnpm prisma migrate dev` to update the schema

## Code Quality & Standards

### TypeScript Configuration
- **Strict mode enabled** with additional safety flags: `noUncheckedIndexedAccess`, `noImplicitReturns`, `noFallthroughCasesInSwitch`
- Path alias: `@/*` maps to `./src/*`
- Target: ES2017 for modern browser support

### Validation & Security
- **API Authentication**: Webhook uses timing-safe comparison (`crypto.timingSafeEqual`) to prevent timing attacks
- **Input Validation**: All API inputs use Zod schemas with detailed error messages
- **SQL Injection Prevention**: Prisma ORM with parameterized queries
- **Data Sanitization**: Markdown content sanitized with `rehype-sanitize` plugin

### Error Handling Patterns
- **Webhook**: Partial success mode - continues processing remaining projects even if individual projects fail
- **Database**: Unique constraints use try-catch with fallback logic (e.g., tag slug conflicts in webhook)
- **Console**: Use `console.warn()` for operational logs, `console.error()` for errors

### Git Commit Conventions
- **Format**: `<type>: <description>` (type in lowercase Chinese: feat/fix/refactor/chore)
- **Types**: `feat` (新功能), `fix` (修复), `refactor` (重构), `chore` (杂项)
- **Examples**:
  - `feat: 新增项目发现任务系统`
  - `fix: 修复 ESLint 警告`
  - `refactor: 重构项目内容标准实现职责分离`

## Testing Strategy

### Unit Tests (Vitest)
- Location: Test files co-located with source code (e.g., `*.test.ts`)
- Run: `pnpm test` for all tests, `pnpm test <pattern>` for specific tests
- Configuration: Vitest with `@testing-library/jest-dom` matchers

### E2E Tests (Playwright)
- Location: `tests/e2e/` or co-located with features
- Run: `pnpm test:e2e` to execute all E2E tests
- Usage: Focus on critical user journeys (project browsing, search, locale switching)

## Performance Considerations

### Database Optimization
- **Index Strategy**: Composite indexes on frequently queried fields (status+createdAt, type+url)
- **N+1 Prevention**: Batch queries for tags (see webhook route:94-98)
- **Connection Pooling**: Prisma client singleton pattern (`src/lib/prisma.ts`)

### Frontend Performance
- **ISR**: Project detail pages revalidated every 5 minutes (`revalidate = 300`)
- **Package Optimization**: Lucide-react imports optimized via `experimental.optimizePackageImports`
- **Image Domains**: Pre-configured for localhost, *.anthropic.com, img.shields.io

### API Rate Limiting
- Webhook: Max 100 projects per request
- Discovery tasks: Max 50 tasks per batch creation
- All queries: Max 100 items per page (enforced via Zod schemas)