feat: 实现 AI 智能搜索功能

添加语义搜索能力，支持自然语言查询找到相关项目。 - 数据库：新增 embedding 字段用于向量存储 - 前端：新增 AI 搜索栏和结果组件，支持传统/AI 模式切换 - API：新增 /api/search/ai 端点处理语义搜索请求 - 国际化：添加 AI 搜索相关中英文翻译 - 探索任务：允许 FAILED 状态直接转到 IN_PROGRESS 简化重试 Co-Authored-By: Claude <noreply@anthropic.com>
2026-01-26 20:59:13 +08:00
parent 6c65cb7cc0
commit 02cbe59a0c
12 changed files with 942 additions and 32 deletions
@@ -0,0 +1,210 @@
+# n8n 工作流导入指南
+
+本指南将帮助您导入和配置 AI 智能搜索系统的两个 n8n 工作流。
+
+## 前置条件
+
+确保您已经完成：
+- [ ] n8n 实例已运行
+- [ ] 已配置 `OpenAI Embeddings` 凭证
+- [ ] 已配置 `Neon Database` 凭证
+- [ ] Neon 数据库已应用迁移（添加 embedding 字段）
+
+---
+
+## 工作流 1: Project Vectorization（项目向量化）
+
+### 功能说明
+每 5 分钟自动执行一次，查询未向量化的项目，生成 OpenAI embeddings 并存储到数据库。
+
+### 导入步骤
+
+1. **导入工作流**
+   - 打开 n8n 实例
+   - 点击右上角 **+** → **Import from File**
+   - 选择 `project-vectorization.json`
+   - 点击 **Import**
+
+2. **配置凭证**
+   - 点击 **查询未向量化项目** 节点
+   - 在 **Credentials** 下拉框中选择 `Neon Database`
+   - 点击 **Save**
+
+   - 点击 **OpenAI Embeddings** 节点
+   - 在 **Credentials** 下拉框中选择 `OpenAI Embeddings`
+   - 点击 **Save**
+
+   - 点击 **更新 Embedding** 节点
+   - 在 **Credentials** 下拉框中选择 `Neon Database`
+   - 点击 **Save**
+
+3. **测试工作流**
+   - 点击工作流右上角 **Test Workflow**
+   - 手动点击 **Cron** 节点的执行按钮
+   - 查看每个节点的输出：
+     - `查询未向量化项目` 应返回项目列表（或空数组）
+     - `构造文本内容` 应添加 `textContent` 字段
+     - `OpenAI Embeddings` 应返回向量数组
+     - `更新 Embedding` 应成功更新数据库
+
+4. **激活工作流**
+   - 点击左上角 **Inactive** 开关，变为 **Active**
+   - 工作流将每 5 分钟自动执行
+
+### 节点说明
+
+| 节点 | 功能 |
+|------|------|
+| Cron | 定时触发器（每 5 分钟） |
+| 查询未向量化项目 | 查询 embedding 为空的 ACTIVE 项目 |
+| 构造文本内容 | 合并项目字段生成用于向量化的文本 |
+| Split in Batches | 分批处理（每批 5 个，避免 API 限流） |
+| OpenAI Embeddings | 调用 OpenAI API 生成向量 |
+| 更新 Embedding | 将向量写入数据库 |
+
+---
+
+## 工作流 2: AI Semantic Search（AI 语义搜索）
+
+### 功能说明
+接收 Webhook 请求，生成查询向量，执行向量相似度搜索，返回排序结果。
+
+### 导入步骤
+
+1. **导入工作流**
+   - 打开 n8n 实例
+   - 点击右上角 **+** → **Import from File**
+   - 选择 `ai-semantic-search.json`
+   - 点击 **Import**
+
+2. **配置凭证**
+   - 依次配置以下节点的凭证为 `OpenAI Embeddings`：
+     - **生成查询向量** 节点
+
+   - 依次配置以下节点的凭证为 `Neon Database`：
+     - **向量相似度搜索** 节点
+     - **查询标签** 节点
+
+3. **获取 Webhook URL**
+   - 点击 **Webhook** 节点
+   - 复制 **Production URL**（格式类似：`https://your-n8n.com/webhook/ai-search`）
+   - 将此 URL 更新到 `.env.local` 的 `N8N_AI_SEARCH_WEBHOOK`
+
+4. **测试工作流**
+   - 点击工作流右上角 **Test Workflow**
+   - 在 **Webhook** 节点中点击 **Listen for Test Event**
+   - 使用以下命令测试：
+
+```bash
+curl -X POST https://your-n8n.com/webhook/ai-search \
+  -H "Content-Type: application/json" \
+  -d '{"query":"视频生成工具","locale":"zh","limit":5}'
+```
+
+   - 预期响应：
+
+```json
+{
+  "results": [
+    {
+      "project": { /* 项目数据 */ },
+      "similarity": 0.89,
+      "matchReason": "相似度: 89%"
+    }
+  ],
+  "total": 5,
+  "searchTime": 1234
+}
+```
+
+5. **激活工作流**
+   - 点击左上角 **Inactive** 开关，变为 **Active**
+
+### 节点说明
+
+| 节点 | 功能 |
+|------|------|
+| Webhook | 接收搜索请求（POST /webhook/ai-search） |
+| 生成查询向量 | 将查询文本转换为向量 |
+| 向量相似度搜索 | 使用 pgvector 执行余弦相似度搜索 |
+| 准备标签查询 | 准备项目 ID 列表 |
+| 查询标签 | 查询每个项目的标签 |
+| 合并标签 | 将标签合并到搜索结果 |
+| 格式化响应 | 生成最终的 JSON 响应 |
+
+---
+
+## 常见问题
+
+### Q1: 节点连接错误？
+导入后如果节点连接线丢失，手动按以下顺序连接：
+
+**工作流 1 连接顺序：**
+```
+Cron → 查询未向量化项目 → 构造文本内容 → Split in Batches → OpenAI Embeddings → 更新 Embedding → (循环回) Split in Batches
+```
+
+**工作流 2 连接顺序：**
+```
+Webhook → 生成查询向量 → 向量相似度搜索 → 准备标签查询 → 查询标签 → 合并标签 → 格式化响应
+```
+
+### Q2: 凭证选择框为空？
+- 确保已在 n8n 中创建了 `OpenAI Embeddings` 和 `Neon Database` 凭证
+- 如果凭证已创建但不可见，重新导入工作流
+
+### Q3: OpenAI API 错误？
+- 检查 API Key 是否有效
+- 确认 API Key 有足够的配额
+- 检查网络连接
+
+### Q4: 数据库连接错误？
+- 验证 Neon 数据库凭证配置正确
+- 检查数据库是否已应用迁移
+- 确认 pgvector 扩展已安装
+
+---
+
+## 更新环境变量
+
+将获取的 Webhook URL 更新到项目的 `.env.local` 文件：
+
+```bash
+# n8n AI Search Webhook
+N8N_AI_SEARCH_WEBHOOK="https://your-n8n.com/webhook/ai-search"
+```
+
+然后重启开发服务器：
+
+```bash
+pnpm dev
+```
+
+---
+
+## 验证完整流程
+
+1. **启动向量化**
+   - 确保 Project Vectorization 工作流已激活
+   - 等待 5 分钟或手动执行
+   - 在 Neon SQL Editor 中检查：
+
+```sql
+SELECT COUNT(*) FROM "projects" WHERE "embedding" IS NOT NULL;
+```
+
+2. **测试 AI 搜索**
+   - 访问 `http://localhost:3000/zh/projects`
+   - 点击 ✨ 按钮切换到 AI 模式
+   - 输入查询："帮我找能生成视频的 AI 工具"
+   - 验证返回相关结果
+
+---
+
+## 完成后
+
+所有工作流配置完成后，您的 AI 智能搜索系统就已就绪！
+
+- 向量化工作流会在后台自动运行
+- AI 搜索 API 可供前端调用
+- 用户可以使用自然语言查询项目
@@ -0,0 +1,180 @@
+{
+  "name": "AI Semantic Search",
+  "nodes": [
+    {
+      "parameters": {
+        "path": "ai-search",
+        "responseMode": "whenLastNodeFinishes",
+        "options": {}
+      },
+      "id": "webhook-node",
+      "name": "Webhook",
+      "type": "n8n-nodes-base.webhook",
+      "typeVersion": 2,
+      "position": [250, 300],
+      "webhookId": "ai-search-webhook"
+    },
+    {
+      "parameters": {
+        "resource": "embedding",
+        "model": "text-embedding-3-small",
+        "input": "={{ $json.query }}",
+        "options": {}
+      },
+      "id": "openai-embeddings",
+      "name": "生成查询向量",
+      "type": "@n8n/n8n-nodes-langchain.openai",
+      "typeVersion": 1.4,
+      "position": [470, 300],
+      "credentials": {
+        "openaiApi": {
+          "id": "OPENAI_CREDENTIAL_ID",
+          "name": "OpenAI Embeddings"
+        }
+      }
+    },
+    {
+      "parameters": {
+        "operation": "executeQuery",
+        "query": "=SELECT\n  p.id,\n  p.name,\n  p.\"nameEn\",\n  p.slug,\n  p.description,\n  p.\"descriptionEn\",\n  p.status,\n  p.\"createdAt\",\n  1 - (p.\"embedding\" <=> '{{ $json.data[0].embedding }}'::vector) as similarity\nFROM \"projects\" p\nWHERE p.\"embedding\" IS NOT NULL\n  AND p.status = 'ACTIVE'\nORDER BY p.\"embedding\" <=> '{{ $json.data[0].embedding }}'::vector\nLIMIT {{ $json.limit || 20 }}",
+        "options": {}
+      },
+      "id": "postgres-vector-search",
+      "name": "向量相似度搜索",
+      "type": "n8n-nodes-base.postgres",
+      "typeVersion": 2.5,
+      "position": [690, 300],
+      "credentials": {
+        "postgres": {
+          "id": "NEON_DATABASE_CREDENTIAL_ID",
+          "name": "Neon Database"
+        }
+      }
+    },
+    {
+      "parameters": {
+        "jsCode": "// 为每个项目查询标签\nconst results = $input.all();\n\nreturn results.map(item => {\n  return {\n    json: {\n      ...item.json,\n      projectId: item.json.id\n    }\n  };\n});"
+      },
+      "id": "prepare-tag-query",
+      "name": "准备标签查询",
+      "type": "n8n-nodes-base.code",
+      "typeVersion": 2,
+      "position": [910, 300]
+    },
+    {
+      "parameters": {
+        "operation": "executeQuery",
+        "query": "=SELECT\n  t.id,\n  t.name,\n  t.\"nameEn\",\n  t.slug,\n  pt.\"projectId\"\nFROM \"tags\" t\nINNER JOIN \"project_tags\" pt ON t.id = pt.\"tagId\"\nWHERE pt.\"projectId\" IN (SELECT UNNEST(STRING_TO_ARRAY('{{ $json.projectIds }}', ','))::INTEGER)",
+        "options": {}
+      },
+      "id": "query-tags",
+      "name": "查询标签",
+      "type": "n8n-nodes-base.postgres",
+      "typeVersion": 2.5,
+      "position": [1130, 300],
+      "credentials": {
+        "postgres": {
+          "id": "NEON_DATABASE_CREDENTIAL_ID",
+          "name": "Neon Database"
+        }
+      }
+    },
+    {
+      "parameters": {
+        "jsCode": "// 合并标签到项目\nconst projects = $('向量相似度搜索').all();\nconst tags = $('查询标签').all();\n\n// 合并标签到项目\nconst results = projects.map(project => {\n  const projectTags = tags\n    .filter(t => t.json.projectId === project.json.id)\n    .map(t => ({\n      id: t.json.id,\n      name: t.json.name,\n      nameEn: t.json.nameEn,\n      slug: t.json.slug\n    }));\n\n  return {\n    json: {\n      project: {\n        ...project.json,\n        tags: projectTags\n      },\n      similarity: project.json.similarity,\n      matchReason: `相似度: ${(project.json.similarity * 100).toFixed(0)}%`\n    }\n  };\n});\n\nreturn results;"
+      },
+      "id": "merge-tags",
+      "name": "合并标签",
+      "type": "n8n-nodes-base.code",
+      "typeVersion": 2,
+      "position": [1350, 300]
+    },
+    {
+      "parameters": {
+        "jsCode": "// 格式化最终响应\nconst results = $input.all();\n\nreturn {\n  json: {\n    results: results.map(r => r.json),\n    total: results.length,\n    searchTime: Date.now() - $('Webhook').item.json.startTime\n  }\n};"
+      },
+      "id": "format-response",
+      "name": "格式化响应",
+      "type": "n8n-nodes-base.code",
+      "typeVersion": 2,
+      "position": [1570, 300]
+    }
+  ],
+  "connections": {
+    "Webhook": {
+      "main": [
+        [
+          {
+            "node": "生成查询向量",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "生成查询向量": {
+      "main": [
+        [
+          {
+            "node": "向量相似度搜索",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "向量相似度搜索": {
+      "main": [
+        [
+          {
+            "node": "准备标签查询",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "准备标签查询": {
+      "main": [
+        [
+          {
+            "node": "查询标签",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "查询标签": {
+      "main": [
+        [
+          {
+            "node": "合并标签",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "合并标签": {
+      "main": [
+        [
+          {
+            "node": "格式化响应",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    }
+  },
+  "pinData": {},
+  "settings": {
+    "executionOrder": "v1"
+  },
+  "staticData": null,
+  "tags": [],
+  "triggerCount": 0,
+  "updatedAt": "2026-01-26T00:00:00.000Z",
+  "versionId": "1"
+}
@@ -0,0 +1,175 @@
+{
+  "name": "Project Vectorization",
+  "nodes": [
+    {
+      "parameters": {
+        "rule": {
+          "interval": [
+            {
+              "field": "minutes",
+              "minutesInterval": 5
+            }
+          ]
+        }
+      },
+      "id": "cron-node",
+      "name": "Cron",
+      "type": "n8n-nodes-base.cron",
+      "typeVersion": 1.2,
+      "position": [250, 300]
+    },
+    {
+      "parameters": {
+        "operation": "executeQuery",
+        "query": "SELECT id, name, \"nameEn\", description, \"descriptionEn\", content, \"contentEn\"\nFROM \"projects\"\nWHERE \"embedding\" IS NULL\n  AND \"status\" = 'ACTIVE'\nLIMIT 20",
+        "options": {}
+      },
+      "id": "postgres-query",
+      "name": "查询未向量化项目",
+      "type": "n8n-nodes-base.postgres",
+      "typeVersion": 2.5,
+      "position": [470, 300],
+      "credentials": {
+        "postgres": {
+          "id": "NEON_DATABASE_CREDENTIAL_ID",
+          "name": "Neon Database"
+        }
+      }
+    },
+    {
+      "parameters": {
+        "jsCode": "// 为每个项目构造用于向量化的文本内容\nconst projects = $input.all();\n\nreturn projects.map(item => {\n  const project = item.json;\n\n  // 合并字段，按权重构造\n  const parts = [\n    project.name || '',\n    project.nameEn || '',\n    project.description || '',\n    project.descriptionEn || '',\n    (project.content || '').substring(0, 500),\n    (project.contentEn || '').substring(0, 500)\n  ].filter(Boolean);\n\n  const textContent = parts.join('\\n\\n');\n\n  return {\n    json: {\n      ...project,\n      textContent: textContent\n    }\n  };\n});"
+      },
+      "id": "construct-content",
+      "name": "构造文本内容",
+      "type": "n8n-nodes-base.code",
+      "typeVersion": 2,
+      "position": [690, 300]
+    },
+    {
+      "parameters": {
+        "batchSize": 5,
+        "options": {}
+      },
+      "id": "split-batches",
+      "name": "Split in Batches",
+      "type": "n8n-nodes-base.splitInBatches",
+      "typeVersion": 3,
+      "position": [910, 300]
+    },
+    {
+      "parameters": {
+        "resource": "embedding",
+        "model": "text-embedding-3-small",
+        "input": "={{ $json.textContent }}",
+        "options": {}
+      },
+      "id": "openai-embeddings",
+      "name": "OpenAI Embeddings",
+      "type": "@n8n/n8n-nodes-langchain.openai",
+      "typeVersion": 1.4,
+      "position": [1130, 300],
+      "credentials": {
+        "openaiApi": {
+          "id": "OPENAI_CREDENTIAL_ID",
+          "name": "OpenAI Embeddings"
+        }
+      }
+    },
+    {
+      "parameters": {
+        "operation": "executeQuery",
+        "query": "=UPDATE \"projects\"\nSET\n  \"embedding\" = '{{ $json.data[0].embedding }}'::vector,\n  \"embeddingUpdatedAt\" = NOW()\nWHERE \"id\" = {{ $json.id }}",
+        "options": {}
+      },
+      "id": "postgres-update",
+      "name": "更新 Embedding",
+      "type": "n8n-nodes-base.postgres",
+      "typeVersion": 2.5,
+      "position": [1350, 300],
+      "credentials": {
+        "postgres": {
+          "id": "NEON_DATABASE_CREDENTIAL_ID",
+          "name": "Neon Database"
+        }
+      }
+    }
+  ],
+  "connections": {
+    "Cron": {
+      "main": [
+        [
+          {
+            "node": "查询未向量化项目",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "查询未向量化项目": {
+      "main": [
+        [
+          {
+            "node": "构造文本内容",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "构造文本内容": {
+      "main": [
+        [
+          {
+            "node": "Split in Batches",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "Split in Batches": {
+      "main": [
+        [
+          {
+            "node": "OpenAI Embeddings",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "OpenAI Embeddings": {
+      "main": [
+        [
+          {
+            "node": "更新 Embedding",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    },
+    "更新 Embedding": {
+      "main": [
+        [
+          {
+            "node": "Split in Batches",
+            "type": "main",
+            "index": 0
+          }
+        ]
+      ]
+    }
+  },
+  "pinData": {},
+  "settings": {
+    "executionOrder": "v1"
+  },
+  "staticData": null,
+  "tags": [],
+  "triggerCount": 0,
+  "updatedAt": "2026-01-26T00:00:00.000Z",
+  "versionId": "1"
+}
@@ -39,23 +39,25 @@ enum TaskStatus {
 // ================================

 model Project {
-  id             String        @id @default(cuid())
-  name           String
-  nameEn         String?
-  slug           String        @unique
-  description    String
-  descriptionEn  String?
-  content        String?       @db.Text
-  contentEn      String?       @db.Text
-  status         ProjectStatus @default(ACTIVE)
-  source         String?
-  createdAt      DateTime      @default(now())
-  updatedAt      DateTime      @updatedAt
+  id                String        @id @default(cuid())
+  name              String
+  nameEn            String?
+  slug              String        @unique
+  description       String
+  descriptionEn     String?
+  content           String?       @db.Text
+  contentEn         String?       @db.Text
+  status            ProjectStatus @default(ACTIVE)
+  source            String?
+  embedding         Unsupported("vector(1536)")?
+  embeddingUpdatedAt DateTime?
+  createdAt         DateTime      @default(now())
+  updatedAt         DateTime      @updatedAt

  // Relations
-  tags           ProjectTag[]
-  links          ExternalLink[]
-  discoveryTasks ProjectDiscoveryTask[]
+  tags              ProjectTag[]
+  links             ExternalLink[]
+  discoveryTasks    ProjectDiscoveryTask[]

  // Indexes
  @@index([status, createdAt], map: "idx_project_status_createdAt")
@@ -0,0 +1,103 @@
+'use client'
+
+import { useState } from 'react'
+import { AISearchBar } from '@/components/search/AISearchBar'
+import { AISearchResults } from '@/components/search/AISearchResults'
+import type { ProjectWithFlatTags } from '@/hooks/useProjects'
+
+interface AISearchResult {
+  project: ProjectWithFlatTags
+  similarity: number
+  matchReason?: string
+}
+
+interface ProjectsPageClientProps {
+  locale: string
+  children: React.ReactNode
+  searchPlaceholder: string
+  searchLabel: string
+  aiPlaceholder: string
+  aiLabel: string
+}
+
+export function ProjectsPageClient({
+  locale,
+  children,
+  searchPlaceholder,
+  searchLabel,
+  aiPlaceholder,
+  aiLabel,
+}: ProjectsPageClientProps) {
+  const [aiResults, setAiResults] = useState<AISearchResult[]>([])
+  const [isAIResult, setIsAIResult] = useState(false)
+  const [loading, setLoading] = useState(false)
+  const [error, setError] = useState<string | null>(null)
+
+  const handleAISearch = async (query: string, isAI: boolean) => {
+    if (!isAI) {
+      // 传统搜索：刷新页面到 URL 参数
+      const params = new URLSearchParams()
+      if (query) params.set('search', query)
+      window.location.href = `/${locale}/projects?${params.toString()}`
+      return
+    }
+
+    // AI 搜索
+    setLoading(true)
+    setError(null)
+    setIsAIResult(true)
+
+    try {
+      const response = await fetch('/api/search/ai', {
+        method: 'POST',
+        headers: { 'Content-Type': 'application/json' },
+        body: JSON.stringify({
+          search: query,
+          locale: locale,
+          limit: 20
+        })
+      })
+
+      if (!response.ok) {
+        throw new Error('搜索失败，请稍后重试')
+      }
+
+      const data = await response.json()
+      setAiResults(data.results || [])
+    } catch (err) {
+      setError(err instanceof Error ? err.message : '未知错误')
+      setAiResults([])
+    } finally {
+      setLoading(false)
+    }
+  }
+
+  return (
+    <>
+      {/* AI 搜索栏 */}
+      <AISearchBar
+        locale={locale}
+        searchPlaceholder={searchPlaceholder}
+        searchLabel={searchLabel}
+        aiPlaceholder={aiPlaceholder}
+        aiLabel={aiLabel}
+        onSearch={handleAISearch}
+        loading={loading}
+      />
+
+      {/* Tag Cloud / AI 搜索结果 */}
+      {isAIResult ? (
+        <>
+          {error && (
+            <div className="mb-8 px-4 py-3 bg-red-50 dark:bg-red-900/20 border-l-4 border-red-500 rounded-r">
+              <p className="text-red-700 dark:text-red-400 font-display text-sm">{error}</p>
+            </div>
+          )}
+          <AISearchResults results={aiResults} locale={locale} />
+        </>
+      ) : (
+        children
+      )}
+    </>
+  )
+}
@@ -3,7 +3,7 @@ import { getTranslations } from 'next-intl/server'
 import { getProjects, getAllTags, getTopTags } from '@/hooks/useProjects'
 import { ProjectList } from '@/components/project/ProjectList'
 import { TagCloud } from '@/components/project/TagCloud'
-import { SearchBar } from '@/components/search/SearchBar'
+import { ProjectsPageClient } from './ProjectsPageClient'

 interface ProjectsPageProps {
  params: Promise<{ locale: string }>
@@ -34,24 +34,27 @@ export default async function ProjectsPage({
      {/* Search and Filter Section */}
      <div className="bg-surface-light dark:bg-surface-dark border-2 border-black dark:border-white/20 p-6 md:p-8 mb-12 shadow-neo dark:shadow-none">
        <Suspense fallback={<div className="h-20"></div>}>
-          <SearchBar
+          <ProjectsPageClient
            locale={locale}
            searchPlaceholder={t('searchPlaceholder')}
            searchLabel={tCommon('search')}
-          />
+            aiPlaceholder={t('aiSearchPlaceholder')}
+            aiLabel={tCommon('aiSearch')}
+          >
+            {/* Tag Cloud - rendered inside client component for non-AI mode */}
+            <div className="border-t-2 border-gray-100 dark:border-gray-800 pt-6 mt-8">
+              <h3 className="font-display font-bold uppercase text-sm mb-4 border-b-2 border-black inline-block dark:border-primary pb-1">
+                Browse by Tags
+              </h3>
+              <TagCloud
+                tags={topTags}      // 默认显示前10个
+                allTags={allTags}   // 用于展开和搜索
+                locale={locale}
+                activeTag={tag}
+              />
+            </div>
+          </ProjectsPageClient>
        </Suspense>
-
-        <div className="border-t-2 border-gray-100 dark:border-gray-800 pt-6 mt-8">
-          <h3 className="font-display font-bold uppercase text-sm mb-4 border-b-2 border-black inline-block dark:border-primary pb-1">
-            Browse by Tags
-          </h3>
-          <TagCloud
-            tags={topTags}      // 默认显示前10个
-            allTags={allTags}   // 用于展开和搜索
-            locale={locale}
-            activeTag={tag}
-          />
-        </div>
      </div>

      {/* Projects Section */}
@@ -8,14 +8,14 @@ import type { Prisma } from '@prisma/client'
 * 有效的任务状态转换规则
 * PENDING -> IN_PROGRESS
 * IN_PROGRESS -> COMPLETED | FAILED
- * FAILED -> PENDING (允许重试)
+ * FAILED -> PENDING | IN_PROGRESS (允许重试，可直接重试或重置后重试)
 * COMPLETED -> (终态，不允许转换)
 */
 const VALID_STATUS_TRANSITIONS: Record<TaskStatus, TaskStatus[]> = {
  PENDING: ['IN_PROGRESS'],
  IN_PROGRESS: ['COMPLETED', 'FAILED'],
  COMPLETED: [],
-  FAILED: ['PENDING'],
+  FAILED: ['PENDING', 'IN_PROGRESS'],
 }

 /**
@@ -0,0 +1,58 @@
+import { NextResponse } from 'next/server'
+import { ZodError } from 'zod'
+import { ProjectQuerySchema } from '@/lib/validations'
+
+const N8N_WEBHOOK_URL = process.env.N8N_AI_SEARCH_WEBHOOK!
+
+if (!N8N_WEBHOOK_URL) {
+  throw new Error('N8N_AI_SEARCH_WEBHOOK environment variable is not set')
+}
+
+export async function POST(request: Request) {
+  try {
+    const body = await request.json()
+
+    // 验证查询参数
+    const validatedQuery = ProjectQuerySchema.parse(body)
+
+    // 转发到 n8n 工作流
+    const n8nResponse = await fetch(N8N_WEBHOOK_URL, {
+      method: 'POST',
+      headers: {
+        'Content-Type': 'application/json',
+      },
+      body: JSON.stringify({
+        query: validatedQuery.search,
+        locale: body.locale || 'zh',
+        limit: validatedQuery.limit || 20,
+        filters: {
+          tags: validatedQuery.tags,
+          status: validatedQuery.status
+        }
+      })
+    })
+
+    if (!n8nResponse.ok) {
+      throw new Error(`n8n webhook failed: ${n8nResponse.statusText}`)
+    }
+
+    const results = await n8nResponse.json()
+
+    return NextResponse.json(results)
+
+  } catch (error) {
+    console.error('AI search error:', error)
+
+    if (error instanceof ZodError) {
+      return NextResponse.json(
+        { error: 'Invalid query parameters', details: error.errors },
+        { status: 400 }
+      )
+    }
+
+    return NextResponse.json(
+      { error: 'AI search failed', message: error instanceof Error ? error.message : 'Unknown error' },
+      { status: 500 }
+    )
+  }
+}
@@ -0,0 +1,96 @@
+'use client'
+
+import { useState } from 'react'
+import { Sparkles } from 'lucide-react'
+
+interface AISearchBarProps {
+  locale: string
+  searchPlaceholder: string
+  searchLabel: string
+  aiPlaceholder: string
+  aiLabel: string
+  onSearch: (query: string, isAI: boolean) => void
+  loading?: boolean
+}
+
+export function AISearchBar({
+  locale,
+  searchPlaceholder,
+  searchLabel,
+  aiPlaceholder,
+  aiLabel,
+  onSearch,
+  loading = false,
+}: AISearchBarProps) {
+  const [query, setQuery] = useState('')
+  const [aiMode, setAiMode] = useState(false)
+
+  const handleSubmit = (e: React.FormEvent) => {
+    e.preventDefault()
+    if (query.trim()) {
+      onSearch(query, aiMode)
+    }
+  }
+
+  const toggleAIMode = () => {
+    setAiMode(!aiMode)
+    setQuery('')
+  }
+
+  return (
+    <form onSubmit={handleSubmit} className="max-w-2xl mx-auto">
+      <div className="relative group">
+        {/* Glow effect on hover */}
+        <div className="absolute -inset-1 bg-black dark:bg-primary rounded-lg blur opacity-25 group-hover:opacity-50 transition duration-200"></div>
+
+        <div className="relative flex items-center gap-2">
+          {/* Search icon */}
+          <div className="absolute inset-y-0 left-0 pl-4 flex items-center pointer-events-none">
+            <span className="text-gray-400">🔍</span>
+          </div>
+
+          {/* Input */}
+          <input
+            type="text"
+            value={query}
+            onChange={(e) => setQuery(e.target.value)}
+            placeholder={aiMode ? aiPlaceholder : searchPlaceholder}
+            className="block w-full pl-12 pr-40 py-4 bg-surface-light dark:bg-surface-dark border-2 border-black dark:border-gray-600 text-text-light dark:text-text-dark placeholder-gray-500 focus:ring-0 focus:border-black dark:focus:border-primary font-display shadow-neo transition-all"
+          />
+
+          {/* AI Mode Toggle */}
+          <button
+            type="button"
+            onClick={toggleAIMode}
+            className={`
+              absolute inset-y-2 right-24 px-3 py-2 font-display font-bold text-sm border-2 transition-all shadow-neo-sm active:shadow-none active:translate-x-[2px] active:translate-y-[2px]
+              ${aiMode
+                ? 'bg-primary text-black border-black hover:bg-yellow-400'
+                : 'bg-white dark:bg-surface-dark text-gray-600 dark:text-gray-400 border-gray-300 dark:border-gray-600 hover:bg-gray-100 dark:hover:bg-white/10'
+              }
+            `}
+            title={aiMode ? '切换到传统搜索' : '切换到 AI 搜索'}
+          >
+            <Sparkles className="w-4 h-4" />
+          </button>
+
+          {/* Search button */}
+          <button
+            type="submit"
+            disabled={loading || !query.trim()}
+            className="absolute inset-y-2 right-2 px-4 bg-black dark:bg-primary text-white dark:text-black font-bold font-display text-sm border-2 border-black dark:border-primary hover:bg-gray-800 dark:hover:bg-yellow-400 disabled:opacity-50 disabled:cursor-not-allowed transition-colors shadow-neo-sm active:shadow-none active:translate-x-[2px] active:translate-y-[2px]"
+          >
+            {loading ? '搜索中...' : (aiMode ? aiLabel : searchLabel)}
+          </button>
+        </div>
+      </div>
+
+      {/* AI Mode Hint */}
+      {aiMode && (
+        <div className="mt-3 text-sm text-gray-600 dark:text-gray-400 font-display">
+          💡 {aiMode ? '试试："帮我找能生成视频的 AI 工具"' : '输入项目名称或描述'}
+        </div>
+      )}
+    </form>
+  )
+}
@@ -0,0 +1,79 @@
+'use client'
+
+import { ProjectCard } from '@/components/project/ProjectCard'
+import type { ProjectWithFlatTags } from '@/hooks/useProjects'
+
+interface AISearchResult {
+  project: ProjectWithFlatTags
+  similarity: number
+  matchReason?: string
+}
+
+interface AISearchResultsProps {
+  results: AISearchResult[]
+  locale: string
+}
+
+export function AISearchResults({ results, locale }: AISearchResultsProps) {
+  if (results.length === 0) {
+    return (
+      <div className="text-center py-12 px-4 bg-surface-light dark:bg-surface-dark border-2 border-dashed border-gray-300 dark:border-gray-700 rounded-lg">
+        <div className="text-4xl mb-4">🔍</div>
+        <p className="text-gray-600 dark:text-gray-400 font-display">
+          未找到相关项目，试试其他描述吧
+        </p>
+      </div>
+    )
+  }
+
+  return (
+    <div className="space-y-6">
+      {/* 相似度说明 */}
+      <div className="flex items-center gap-3 px-4 py-3 bg-gray-50 dark:bg-gray-900 border-l-4 border-primary rounded-r font-display text-sm">
+        <span className="font-semibold text-gray-700 dark:text-gray-300">匹配度：</span>
+        <div className="flex items-center gap-2">
+          <span className="text-green-600 font-bold">高</span>
+          <span className="text-gray-400">→</span>
+          <span className="text-red-600 font-bold">低</span>
+        </div>
+      </div>
+
+      {/* 结果列表 */}
+      {results.map(({ project, similarity, matchReason }) => (
+        <div key={project.id} className="relative">
+          {/* 相似度指示条 */}
+          <div
+            className="absolute left-0 top-0 bottom-0 w-1.5 rounded-l"
+            style={{
+              backgroundColor: getSimilarityColor(similarity)
+            }}
+          />
+
+          {/* 项目卡片 */}
+          <div className="ml-3">
+            <ProjectCard project={project} locale={locale} />
+
+            {/* AI 匹配信息 */}
+            <div className="mt-3 px-4 py-3 bg-yellow-50 dark:bg-yellow-900/20 border-l-4 border-yellow-400 dark:border-yellow-500 rounded-r">
+              <div className="flex flex-col sm:flex-row justify-between items-start sm:items-center gap-2 text-sm font-display">
+                <span className="font-semibold text-gray-700 dark:text-gray-300">
+                  匹配度: {(similarity * 100).toFixed(0)}%
+                </span>
+                {matchReason && (
+                  <span className="text-gray-600 dark:text-gray-400">{matchReason}</span>
+                )}
+              </div>
+            </div>
+          </div>
+        </div>
+      ))}
+    </div>
+  )
+}
+
+function getSimilarityColor(score: number): string {
+  if (score > 0.8) return '#22c55e' // green-500
+  if (score > 0.6) return '#eab308' // yellow-500
+  if (score > 0.4) return '#f97316' // orange-500
+  return '#ef4444' // red-500
+}
@@ -1,6 +1,7 @@
 {
  "common": {
    "search": "Search",
+    "aiSearch": "AI Search",
    "loading": "Loading...",
    "noResults": "No results found",
    "noProjects": "No projects yet",
@@ -21,6 +22,7 @@
    "featuredProjects": "Featured Projects",
    "browseByTag": "Browse by Tag",
    "searchPlaceholder": "Search AI projects...",
+    "aiSearchPlaceholder": "Describe what you're looking for, e.g.: AI tools that can generate videos...",
    "metaTitle": "Agent Park - AI Project Navigator",
    "metaDescription": "Discover and explore quality AI projects from across the web",
    "heroTitle": "AI PROJECT",
@@ -1,6 +1,7 @@
 {
  "common": {
    "search": "搜索",
+    "aiSearch": "AI 搜索",
    "loading": "加载中...",
    "noResults": "未找到结果",
    "noProjects": "暂无项目",
@@ -21,6 +22,7 @@
    "featuredProjects": "精选项目",
    "browseByTag": "按标签浏览",
    "searchPlaceholder": "搜索 AI 项目...",
+    "aiSearchPlaceholder": "描述你想要的项目，如：能生成视频的 AI 工具...",
    "metaTitle": "Agent Park - AI 项目导航",
    "metaDescription": "发现和探索全网优质 AI 项目",
    "heroTitle": "AI 项目",