基于AI的Google Sheets网页研究（GPT与Bright Data）

• •HTTP Webhook 端点（n8n 会自动生成）
• •可能需要目标 API 的认证凭证
• •OpenAI API Key

{
  "meta": {
    "instanceId": "ca49a649cff334d4dc2badc0109f1718ecfce65b8ecc6e46df324506dd844436",
    "templateCredsSetupCompleted": true
  },
  "nodes": [
    {
      "id": "14cd7efc-ef41-418a-af6d-bedef22b8085",
      "name": "响应 Webhook",
      "type": "n8n-nodes-base.respondToWebhook",
      "position": [
        -480,
        304
      ],
      "parameters": {
        "options": {
          "responseHeaders": {
            "entries": [
              {
                "name": "Content-Type",
                "value": "text/plain; charset=utf-8"
              }
            ]
          }
        },
        "respondWith": "text",
        "responseBody": "={{ $json.output.summary }}"
      },
      "typeVersion": 1.4
    },
    {
      "id": "b33b25ad-aff9-4192-946b-becd709dfe6c",
      "name": "Bright Data 搜索代理",
      "type": "@n8n/n8n-nodes-langchain.agent",
      "position": [
        -592,
        -96
      ],
      "parameters": {
        "text": "=You are an AI agent specialized in web search and information retrieval.\n\nQuery: \"{{ $json.output }}\"\n\n## Objective:  \nPerform a focused search and return the most relevant link in **JSON format only**:  \n```json\n{\n  \"link\": \"\"\n}\n\n## Execution Process:\n\n### 1. Run Search\nsearch_engine(\n  query: [the query you constructed],\n  engine: \"google\",\n  limit: 10,\n  country: \"us\" or \"il\" depending on relevance\n)\n\n### 2. Filter Results\n\nPrioritize reliable sources:\nMajor news websites (Reuters, Bloomberg, WSJ, Haaretz, Globes)\nOfficial websites (investor relations, about pages)\nOfficial financial reports (SEC filings, quarterly reports)\nTrusted data sources (Wikipedia, Crunchbase, LinkedIn)\n\nSkip:\nAds and irrelevant marketing content\nUnverified forums\nOutdated content (unless historical info is requested)\nBroken or invalid links\n\nFind a suitable link based on:\nRelevance to the query\nRecency of the source\nSource credibility\n\n### 3. Return Result\n\nRequired format – JSON only, no additional text:\n{\n  \"link\": \"https://...\"\n}\n\nIf no high-quality result is found, search another source until you find one or return \"\".\n\n## Error Handling:\n\nIf search fails → try an alternative query\nIf no results found → return JSON with empty fields\nAlways return valid JSON, with no explanations or extra text",
        "options": {},
        "promptType": "define",
        "hasOutputParser": true
      },
      "typeVersion": 2.2
    },
    {
      "id": "d018693a-0fdf-4ae8-ae22-1b4f79bed2a1",
      "name": "Bright Data MCP",
      "type": "@n8n/n8n-nodes-langchain.mcpClientTool",
      "position": [
        -528,
        64
      ],
      "parameters": {
        "include": "selected",
        "options": {
          "timeout": 120000
        },
        "endpointUrl": "https://mcp.brightdata.com/mcp?token=57b289e26c7442f93e806ffe57409116f831983859d456663c260c14a9097ef0&pro=1",
        "includeTools": [
          "search_engine"
        ],
        "serverTransport": "httpStreamable"
      },
      "typeVersion": 1.1
    },
    {
      "id": "ddf8a398-38ae-4da4-a2aa-0fe6a4b3e62b",
      "name": "Webhook 调用",
      "type": "n8n-nodes-base.webhook",
      "position": [
        -1664,
        -32
      ],
      "webhookId": "adc93cdb-28a7-4e48-ad3e-b5c8924f92fa",
      "parameters": {
        "path": "brightdata-search",
        "options": {},
        "httpMethod": "POST",
        "responseMode": "responseNode",
        "authentication": "headerAuth"
      },
      "credentials": {
        "httpHeaderAuth": {
          "id": "D5oQ7Gi9TGPHo4oc",
          "name": "82Labs - Webhook"
        }
      },
      "typeVersion": 2.1
    },
    {
      "id": "8dfdd46c-ad8f-458a-a403-45910dcc13e5",
      "name": "调整查询代理",
      "type": "@n8n/n8n-nodes-langchain.agent",
      "position": [
        -1120,
        -96
      ],
      "parameters": {
        "text": "=User prompt: {{ $json.userPrompt }}\nPrompt's referral: {{ $json.cellReference }}\n",
        "options": {
          "systemMessage": "=### 1. Request Analysis  \n\nIdentify the query type:  \n- **News/Updates**: \"News about X\", \"What’s happening with X\"  \n- **Financial Data**: \"Revenue\", \"Earnings\", \"Financial Reports\" \n- **Factual Information**: \"Who is the CEO\", \"How many employees\", \"Where is the HQ\"  \n- **Analysis/Comparison**: \"Competitors\", \"Compare to Y\", \"Alternatives\"  \n- **General Research**: Open-ended questions about a topic/field  \n\n### 2. Search Query Construction  \nBuild an optimal query based on the type:\n\n**News**: `\"{{ entity }}\" news [year/month if relevant]`  \n**Financial Data**: `\"{{ entity }}\" revenue earnings \"Q[X] [year]\"` or `financial results`  \n**Factual Information**: `\"{{ entity }}\" [specific fact] official`  \n**Analysis**: `\"{{ entity }}\" analysis competitors market share`  \n\n**Principles:**  \n- Always translate to English to maximize results  \n- Keep double quotes around specific names  \n- Add relevant time — current time is {{ $now.format('dd/LL/yyyy') }}  \n- Match keywords to the requested information type  \n\n## Examples:\n\n**Input**: Prompt's referral=\"Apple\", user prompt=\"Search for news about the company\"  \noutput example: \"Apple Inc\" news 2025 technology  \n\n**Input**: Prompt's referral=\"Tesla\", user prompt=\"Monthly revenue April\"  \noutput example: \"Tesla\" revenue \"April 2024\" OR \"Q2 2024\" financial results  \n\n**Input**: Prompt's referral=\"United States\", user prompt=\"Who is the president\"  \noutput example: \"United States\" president 19/10/2025 current  \n"
        },
        "promptType": "define"
      },
      "typeVersion": 2.2
    },
    {
      "id": "c12218ae-674c-4bf9-b35b-6a8d6b360c9a",
      "name": "Bright Data - 数据提取",
      "type": "n8n-nodes-base.httpRequest",
      "position": [
        -192,
        -48
      ],
      "parameters": {
        "url": "https://api.brightdata.com/request",
        "method": "POST",
        "options": {
          "batching": {
            "batch": {
              "batchSize": 1,
              "batchInterval": 2000
            }
          }
        },
        "sendBody": true,
        "sendHeaders": true,
        "bodyParameters": {
          "parameters": [
            {
              "name": "zone",
              "value": "mcp_unlocker"
            },
            {
              "name": "url",
              "value": "={{ $json.output.link }}"
            },
            {
              "name": "format",
              "value": "json"
            },
            {
              "name": "method",
              "value": "GET"
            },
            {
              "name": "country",
              "value": "il"
            },
            {
              "name": "data_format",
              "value": "markdown"
            }
          ]
        },
        "headerParameters": {
          "parameters": [
            {
              "name": "Authorization",
              "value": "Bearer YOUR_TOKEN_HERE"
            }
          ]
        }
      },
      "typeVersion": 4.2
    },
    {
      "id": "9f8b11b7-f997-4906-a929-2656cdee4204",
      "name": "提取数据",
      "type": "@n8n/n8n-nodes-langchain.chainLlm",
      "onError": "continueRegularOutput",
      "position": [
        -1600,
        336
      ],
      "parameters": {
        "text": "=## Input\n### The user's original request:\n{{ $('Set Variables').item.json.cellReference }} - {{ $('Set Variables').item.json.userPrompt }}\n### Full content scanned from a website:\n{{ $json.body }}",
        "batching": {},
        "messages": {
          "messageValues": [
            {
              "message": "=# You are an AI agent specialized in extracting relevant information. Your role is to receive:\n\n1. **The user's original request** – the question or topic requested  \n2. **Full content scanned from a website** – raw information from a source  \n\n## Objective\n\nExtract and summarize the most relevant information from the scanned content based on the user's query.\n\n## Workflow\n\n1. Carefully read the scanned content  \n2. Identify information directly relevant to the user's query  \n3. Create a focused and concise summary of only the relevant information  \n4. Completely ignore unrelated content  \n\n## Output Rules\n\n- **Return only valid JSON** according to the schema below  \n- Do not include any text outside the JSON  \n- The summary should include **only** information directly relevant to the user's query  \n- Maximum preferred length: **400 characters**  \n- If no relevant information is found, leave the `summary` field empty (`\"\"`)  \n\n## Edge Case Handling\n\n- **Partial or corrupted content:** Extract only the existing and relevant information  \n- **Multiple topics in content:** Focus only on the topic relevant to the user's query  \n- **Different language:** Translate relevant information to {{ $('Set Variables').item.json.ouputLanguage }}  \n- **No relevant information:** Return JSON with an empty `summary`  \n- **Always return valid JSON**, no extra text  \n\n## Key Principles\n\n** Extract factual, accurate, and focused information **  \n** Translate to {{ $('Set Variables').item.json.ouputLanguage }} when needed **  \n** Summarize concisely within 400 characters **  \n\n** Do not add information that is not in the original content **  \n** Do not include information unrelated to the query **  \n** Do not deviate from the JSON format **  \n"
            }
          ]
        },
        "promptType": "define",
        "hasOutputParser": true
      },
      "typeVersion": 1.7
    },
    {
      "id": "d7a349a0-0158-4ec2-bb94-beebebcbe2bd",
      "name": "汇总信息",
      "type": "@n8n/n8n-nodes-langchain.agent",
      "position": [
        -1120,
        336
      ],
      "parameters": {
        "text": "=scraping summary information: {{ $json.output.summary }}\nthe actual user request/question: {{ $('Set Variables').item.json.cellReference }} - {{ $('Set Variables').item.json.userPrompt }}\n",
        "options": {
          "systemMessage": "=Generate a conclusion in {{ $('Set Variables').item.json.ouputLanguage }} based on all the information above. Filter and summarize the most relevant information according to the user's question into a maximum of 400 characters.  \nAlways return valid JSON, with no explanations or additional text.  \n"
        },
        "promptType": "define",
        "hasOutputParser": true
      },
      "typeVersion": 2.2
    },
    {
      "id": "4ca74428-5c4a-4c3f-8359-95a1df48a4b9",
      "name": "更新日志",
      "type": "n8n-nodes-base.dataTable",
      "position": [
        -480,
        496
      ],
      "parameters": {
        "columns": {
          "value": {
            "output": "={{ $json.output.summary }}",
            "input_prompt": "={{ $('Set Variables').item.json.userPrompt }} - {{ $('Set Variables').item.json.cellReference }}"
          },
          "schema": [
            {
              "id": "input_prompt",
              "type": "string",
              "display": true,
              "removed": false,
              "readOnly": false,
              "required": false,
              "displayName": "input_prompt",
              "defaultMatch": false
            },
            {
              "id": "output",
              "type": "string",
              "display": true,
              "removed": false,
              "readOnly": false,
              "required": false,
              "displayName": "output",
              "defaultMatch": false
            }
          ],
          "mappingMode": "defineBelow",
          "matchingColumns": [
            "logs"
          ],
          "attemptToConvertTypes": false,
          "convertFieldsToString": false
        },
        "options": {},
        "dataTableId": {
          "__rl": true,
          "mode": "list",
          "value": "vAMxTuckOJDPlMN3",
          "cachedResultUrl": "/projects/qoZIbTB1W6TCihY8/datatables/vAMxTuckOJDPlMN3",
          "cachedResultName": "BrightData Test"
        }
      },
      "typeVersion": 1
    },
    {
      "id": "1abd9a03-bbb3-44cf-b23c-40363d54458e",
      "name": "结构化输出解析器 - 1",
      "type": "@n8n/n8n-nodes-langchain.outputParserStructured",
      "position": [
        -352,
        64
      ],
      "parameters": {
        "jsonSchemaExample": "{\n  \"link\": \"\"\n}"
      },
      "typeVersion": 1.3
    },
    {
      "id": "0b99dc9d-bc59-4054-b97a-7bc708439a22",
      "name": "结构化输出解析器 - 2",
      "type": "@n8n/n8n-nodes-langchain.outputParserStructured",
      "position": [
        -1392,
        496
      ],
      "parameters": {
        "jsonSchemaExample": "{\n  \"summary\": \"\"\n}"
      },
      "typeVersion": 1.3
    },
    {
      "id": "1987c731-3ef5-4977-8029-8415726355ea",
      "name": "GPT 4o Mini - 1",
      "type": "@n8n/n8n-nodes-langchain.lmChatOpenAi",
      "position": [
        -1664,
        496
      ],
      "parameters": {
        "model": {
          "__rl": true,
          "mode": "list",
          "value": "gpt-4o-mini",
          "cachedResultName": "gpt-4o-mini"
        },
        "options": {}
      },
      "credentials": {
        "openAiApi": {
          "id": "KFAwjhWN2Hk6Vvhy",
          "name": "General - OpenAi"
        }
      },
      "typeVersion": 1.2
    },
    {
      "id": "2c9c2b61-56f7-40ef-912c-1a1bf34688a2",
      "name": "GPT 4o Mini - 2",
      "type": "@n8n/n8n-nodes-langchain.lmChatOpenAi",
      "position": [
        -1184,
        496
      ],
      "parameters": {
        "model": {
          "__rl": true,
          "mode": "list",
          "value": "gpt-4o-mini",
          "cachedResultName": "gpt-4o-mini"
        },
        "options": {}
      },
      "credentials": {
        "openAiApi": {
          "id": "KFAwjhWN2Hk6Vvhy",
          "name": "General - OpenAi"
        }
      },
      "typeVersion": 1.2
    },
    {
      "id": "7fd1cabe-0429-4ca4-81ba-8b2e6344279b",
      "name": "GPT 4o - 1",
      "type": "@n8n/n8n-nodes-langchain.lmChatOpenAi",
      "position": [
        -704,
        64
      ],
      "parameters": {
        "model": {
          "__rl": true,
          "mode": "list",
          "value": "gpt-4o",
          "cachedResultName": "gpt-4o"
        },
        "options": {}
      },
      "credentials": {
        "openAiApi": {
          "id": "KFAwjhWN2Hk6Vvhy",
          "name": "General - OpenAi"
        }
      },
      "typeVersion": 1.2
    },
    {
      "id": "95e48871-e554-442c-83fb-b028a020b342",
      "name": "GPT 4.1 Mini - 1",
      "type": "@n8n/n8n-nodes-langchain.lmChatOpenAi",
      "position": [
        -1184,
        80
      ],
      "parameters": {
        "model": {
          "__rl": true,
          "mode": "list",
          "value": "gpt-4.1-mini"
        },
        "options": {}
      },
      "credentials": {
        "openAiApi": {
          "id": "KFAwjhWN2Hk6Vvhy",
          "name": "General - OpenAi"
        }
      },
      "typeVersion": 1.2
    },
    {
      "id": "759b9362-6511-47a5-b96c-a7cb340bae16",
      "name": "结构化输出解析器 - 3",
      "type": "@n8n/n8n-nodes-langchain.outputParserStructured",
      "position": [
        -912,
        496
      ],
      "parameters": {
        "jsonSchemaExample": "{\n\t\"summary\": \"Intel was founded in 1968.\"\n}\n"
      },
      "typeVersion": 1.3
    },
    {
      "id": "3b9a3797-81bd-4120-9c36-4d5d2c21ac01",
      "name": "AI 驱动的网页抓取到 Google Sheets 与 Bright Data",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        -2160,
        -208
      ],
      "parameters": {
        "color": 6,
        "width": 396,
        "height": 860,
        "content": "# AI 驱动的网页抓取到 Google Sheets 与 Bright Data"
      },
      "typeVersion": 1
    },
    {
      "id": "14a7088b-82e9-41b4-acbe-899b5eee195b",
      "name": "便签22",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        -1744,
        -208
      ],
      "parameters": {
        "color": 5,
        "width": 500,
        "height": 420,
        "content": "# 1) 获取查询并组织数据"
      },
      "typeVersion": 1
    },
    {
      "id": "97646605-6f57-49f1-9bd6-0afcf176bbe8",
      "name": "便签20",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        -1232,
        -208
      ],
      "parameters": {
        "color": 2,
        "width": 468,
        "height": 420,
        "content": "# 2) 调整查询以适应 Brightdata"
      },
      "typeVersion": 1
    },
    {
      "id": "41e0d6fc-77ce-4541-8522-47f72d9d7782",
      "name": "便签1",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        -752,
        -208
      ],
      "parameters": {
        "color": 4,
        "width": 724,
        "height": 420,
        "content": "# 3) Brightdata 的抓取"
      },
      "typeVersion": 1
    },
    {
      "id": "40feaec4-3243-4c39-be4a-dbd9430707c8",
      "name": "便签15",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        -1744,
        224
      ],
      "parameters": {
        "color": 3,
        "width": 1716,
        "height": 424,
        "content": "# 4) 信息提取 + 检索摘要和更新日志"
      },
      "typeVersion": 1
    },
    {
      "id": "f6fd6c4c-24b2-4e9f-a45d-9e84af36b0dd",
      "name": "设置变量",
      "type": "n8n-nodes-base.set",
      "position": [
        -1424,
        -32
      ],
      "parameters": {
        "options": {},
        "assignments": {
          "assignments": [
            {
              "id": "a4357b33-04b4-4246-ad30-f0d91b6687d2",
              "name": "userPrompt",
              "type": "string",
              "value": "={{ $json.body.source }}"
            },
            {
              "id": "977e4bde-3030-4d02-9c2f-3d46975901be",
              "name": "cellReference",
              "type": "string",
              "value": "={{ $json.body.prompt }}"
            },
            {
              "id": "c63d7766-11b9-4edb-92db-90e25535721b",
              "name": "ouputLanguage",
              "type": "string",
              "value": "Hebrew"
            }
          ]
        }
      },
      "typeVersion": 3.4
    }
  ],
  "pinData": {},
  "connections": {
    "GPT 4o - 1": {
      "ai_languageModel": [
        [
          {
            "node": "Bright Data Search Agent",
            "type": "ai_languageModel",
            "index": 0
          }
        ]
      ]
    },
    "Extract Data": {
      "main": [
        [
          {
            "node": "Summarize Information",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Webhook Call": {
      "main": [
        [
          {
            "node": "Set Variables",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Set Variables": {
      "main": [
        [
          {
            "node": "Adjust Query Agent",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Bright Data MCP": {
      "ai_tool": [
        [
          {
            "node": "Bright Data Search Agent",
            "type": "ai_tool",
            "index": 0
          }
        ]
      ]
    },
    "GPT 4o Mini - 1": {
      "ai_languageModel": [
        [
          {
            "node": "Extract Data",
            "type": "ai_languageModel",
            "index": 0
          }
        ]
      ]
    },
    "GPT 4o Mini - 2": {
      "ai_languageModel": [
        [
          {
            "node": "Summarize Information",
            "type": "ai_languageModel",
            "index": 0
          }
        ]
      ]
    },
    "GPT 4.1 Mini - 1": {
      "ai_languageModel": [
        [
          {
            "node": "Adjust Query Agent",
            "type": "ai_languageModel",
            "index": 0
          }
        ]
      ]
    },
    "Adjust Query Agent": {
      "main": [
        [
          {
            "node": "Bright Data Search Agent",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Summarize Information": {
      "main": [
        [
          {
            "node": "Respond to Webhook",
            "type": "main",
            "index": 0
          },
          {
            "node": "Update Logs",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Bright Data Search Agent": {
      "main": [
        [
          {
            "node": "Bright Data - Data Extraction",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Structured Output Parser - 1": {
      "ai_outputParser": [
        [
          {
            "node": "Bright Data Search Agent",
            "type": "ai_outputParser",
            "index": 0
          }
        ]
      ]
    },
    "Structured Output Parser - 2": {
      "ai_outputParser": [
        [
          {
            "node": "Extract Data",
            "type": "ai_outputParser",
            "index": 0
          }
        ]
      ]
    },
    "Structured Output Parser - 3": {
      "ai_outputParser": [
        [
          {
            "node": "Summarize Information",
            "type": "ai_outputParser",
            "index": 0
          }
        ]
      ]
    },
    "Bright Data - Data Extraction": {
      "main": [
        [
          {
            "node": "Extract Data",
            "type": "main",
            "index": 0
          }
        ]
      ]
    }
  }
}