spy tool
工作流概述
这是一个包含14个节点的复杂工作流,主要用于自动化处理各种任务。
工作流源代码
{
"id": "aOP0D1cAqzGv7Xa8",
"meta": {
"instanceId": "0a5638e14e0c728ef975d18d109cfb41edae575e3d911724f4f1eccde06a729f"
},
"name": "spy tool",
"tags": [],
"nodes": [
{
"id": "5690844d-5322-4c62-8c83-eb4d4dc9c481",
"name": "OpenAI Chat Model",
"type": "@n8n/n8n-nodes-langchain.lmChatOpenAi",
"position": [
1400,
340
],
"parameters": {
"model": "gpt-4o",
"options": {}
},
"credentials": {
"openAiApi": {
"id": "ZOKbogCxHnP2W0H5",
"name": "OpenAi account"
}
},
"typeVersion": 1
},
{
"id": "3b1c034f-501b-423c-844f-9cb607fa91e6",
"name": "Gmail",
"type": "n8n-nodes-base.gmailTool",
"position": [
1580,
340
],
"webhookId": "6a510528-22e0-4140-b987-770bb7a138de",
"parameters": {
"sendTo": "tom@sleak.chat",
"message": "={{ $fromAI(\"change\", \"What relevant part has changed on the website?\") }}",
"options": {
"appendAttribution": false
},
"subject": "=Relevant changes on {{ $('parse results').item.json.website_url }}",
"emailType": "text",
"descriptionType": "manual",
"toolDescription": "=Use this tool if you need to send an email, but only if the terms in the instructions mentioned explicitly state so
"
},
"credentials": {
"gmailOAuth2": {
"id": "jtANm6k92Kl6ent1",
"name": "Gmail account"
}
},
"typeVersion": 2.1
},
{
"id": "4d448a02-4569-451e-8be5-59bfc48f36d8",
"name": "parse results",
"type": "n8n-nodes-base.code",
"position": [
1180,
-160
],
"parameters": {
"jsCode": "const parsedObject = JSON.parse($('convert message to website url & instruction').first().json.choices[0].message.content);
return parsedObject"
},
"typeVersion": 2
},
{
"id": "238298c4-5bba-4ac1-b3cc-ab5a28888560",
"name": "Sticky Note",
"type": "n8n-nodes-base.stickyNote",
"position": [
1420,
-120
],
"parameters": {
"width": 260,
"height": 180,
"content": "## Note: almost never works right away
Adjust the prompts in the 'Tools agent' and 'Gmail' node as desired to steer the agent's behavior in the right direction"
},
"typeVersion": 1
},
{
"id": "0d519c06-aa30-4a33-895f-9185936d27cf",
"name": "Sticky Note2",
"type": "n8n-nodes-base.stickyNote",
"position": [
480,
100
],
"parameters": {
"width": 150,
"height": 80,
"content": "Connect your Firecrawl account"
},
"typeVersion": 1
},
{
"id": "9e327bbe-0096-4a4d-aec2-2e4cae7d91bd",
"name": "Sticky Note3",
"type": "n8n-nodes-base.stickyNote",
"position": [
1740,
80
],
"parameters": {
"width": 150,
"height": 80,
"content": "Connect your own OpenAI account
"
},
"typeVersion": 1
},
{
"id": "30ce0e22-f536-462f-8f94-f3fd92ae036f",
"name": "Sticky Note4",
"type": "n8n-nodes-base.stickyNote",
"position": [
1660,
300
],
"parameters": {
"width": 150,
"height": 80,
"content": "Connect your own Gmail account
"
},
"typeVersion": 1
},
{
"id": "bc003781-3d91-49b6-b6bb-b2970b39256a",
"name": "convert message to website url & instruction",
"type": "n8n-nodes-base.httpRequest",
"position": [
940,
-160
],
"parameters": {
"url": "https://api.openai.com/v1/chat/completions",
"method": "POST",
"options": {},
"jsonBody": "={
\"model\": \"gpt-4o-2024-08-06\",
\"messages\": [
{
\"role\": \"user\",
\"content\": \"convert the following message to a website url (just the plain text url, NOT formatted or in markdown) and prompt to AI. Make the prompt as verbose as possible. Message: {{ $('New espionage assignment').first().json.assignment_instructions }}\"
}
],
\"response_format\": {
\"type\": \"json_schema\",
\"json_schema\": {
\"name\": \"variable_extraction\",
\"schema\": {
\"type\": \"object\",
\"properties\": {
\"website_url\": { \"type\": \"string\" },
\"prompt\": { \"type\": \"string\" }
},
\"required\": [\"website_url\", \"prompt\"],
\"additionalProperties\": false
},
\"strict\": true
}
}
}
",
"sendBody": true,
"specifyBody": "json",
"authentication": "predefinedCredentialType",
"nodeCredentialType": "openAiApi"
},
"credentials": {
"openAiApi": {
"id": "ZOKbogCxHnP2W0H5",
"name": "OpenAi account"
}
},
"typeVersion": 4.2
},
{
"id": "6a8c172d-ac39-4cb0-b601-39fc770695ed",
"name": "New espionage assignment",
"type": "n8n-nodes-base.formTrigger",
"position": [
700,
-160
],
"webhookId": "7470334f-93e1-47af-9521-d3a232c38b13",
"parameters": {
"options": {},
"formTitle": "New espionage assignment",
"formFields": {
"values": [
{
"fieldLabel": "assignment_instructions"
}
]
}
},
"notesInFlow": false,
"typeVersion": 2.2
},
{
"id": "c5c64e5c-88de-45e3-bb9b-4096e74a6e83",
"name": "wait 1 day",
"type": "n8n-nodes-base.wait",
"position": [
940,
80
],
"webhookId": "22e689e4-b93d-4c59-81e5-43c070833454",
"parameters": {
"unit": "days",
"amount": 1
},
"typeVersion": 1.1
},
{
"id": "62a278ff-ed00-4e54-a608-001237551113",
"name": "scrape page - 1",
"type": "n8n-nodes-base.httpRequest",
"position": [
700,
80
],
"parameters": {
"url": "https://api.firecrawl.dev/v1/scrape",
"method": "POST",
"options": {},
"jsonBody": "={
\"url\": \"{{ $('parse results').item.json.website_url }}\",
\"formats\": [
\"markdown\"
],
\"onlyMainContent\": true,
\"waitFor\": 5000
}",
"sendBody": true,
"specifyBody": "json",
"authentication": "genericCredentialType",
"genericAuthType": "httpHeaderAuth"
},
"credentials": {
"httpBasicAuth": {
"id": "h2XRcXzLcEfvDVKb",
"name": "Unnamed credential"
},
"httpHeaderAuth": {
"id": "FoyIka0WgFG4FPxA",
"name": "Header Auth account 2"
}
},
"retryOnFail": true,
"typeVersion": 4.2
},
{
"id": "89c15d8f-7f8e-4391-b24a-07579964ca5c",
"name": "scrape page - 2",
"type": "n8n-nodes-base.httpRequest",
"position": [
1180,
80
],
"parameters": {
"url": "https://api.firecrawl.dev/v1/scrape",
"method": "POST",
"options": {},
"jsonBody": "={
\"url\": \"{{ $('parse results').item.json.website_url }}\",
\"formats\": [
\"markdown\"
],
\"onlyMainContent\": true,
\"waitFor\": 5000
}",
"sendBody": true,
"specifyBody": "json",
"authentication": "genericCredentialType",
"genericAuthType": "httpHeaderAuth"
},
"credentials": {
"httpHeaderAuth": {
"id": "FoyIka0WgFG4FPxA",
"name": "Header Auth account 2"
}
},
"retryOnFail": true,
"typeVersion": 4.2
},
{
"id": "7b148c5b-d4ae-498a-b7ef-2ed4ecc0a665",
"name": "send e-mail?",
"type": "@n8n/n8n-nodes-langchain.agent",
"position": [
1420,
80
],
"parameters": {
"text": "={{ $('parse results').item.json.prompt }}
NOTE: ONLY send an email if the situation meets the above condition. Otherwise, do NOT use the tool
NOTE: this concerns differences between the \"old version page\" (scrape from yesterday) and \"new version page\" (scrape from now)",
"options": {
"systemMessage": "=old version page: \n\n {{ JSON.stringify($('scrape page - 1').item.json[\"data\"][\"markdown\"]) }} \n\n /// \n\n new version page: \n\n {{ JSON.stringify($('scrape page - 1').item.json[\"data\"][\"markdown\"]) }}"
},
"promptType": "define",
"hasOutputParser": true
},
"typeVersion": 1.7
},
{
"id": "7897d707-2c27-43bf-9ea0-90ab7996bf4a",
"name": "Sticky Note5",
"type": "n8n-nodes-base.stickyNote",
"position": [
920,
-260
],
"parameters": {
"width": 150,
"height": 80,
"content": "Connect your own OpenAI account
"
},
"typeVersion": 1
}
],
"active": true,
"pinData": {},
"settings": {
"executionOrder": "v1"
},
"versionId": "dec23eea-1590-4418-ab2b-1cb4a6ccfdc6",
"connections": {
"Gmail": {
"ai_tool": [
[
{
"node": "send e-mail?",
"type": "ai_tool",
"index": 0
}
]
]
},
"wait 1 day": {
"main": [
[
{
"node": "scrape page - 2",
"type": "main",
"index": 0
}
]
]
},
"send e-mail?": {
"main": [
[
{
"node": "scrape page - 1",
"type": "main",
"index": 0
}
]
]
},
"parse results": {
"main": [
[
{
"node": "scrape page - 1",
"type": "main",
"index": 0
}
]
]
},
"scrape page - 1": {
"main": [
[
{
"node": "wait 1 day",
"type": "main",
"index": 0
}
]
]
},
"scrape page - 2": {
"main": [
[
{
"node": "send e-mail?",
"type": "main",
"index": 0
}
]
]
},
"OpenAI Chat Model": {
"ai_languageModel": [
[
{
"node": "send e-mail?",
"type": "ai_languageModel",
"index": 0
}
]
]
},
"New espionage assignment": {
"main": [
[
{
"node": "convert message to website url & instruction",
"type": "main",
"index": 0
}
]
]
},
"convert message to website url & instruction": {
"main": [
[
{
"node": "parse results",
"type": "main",
"index": 0
}
]
]
}
}
}
功能特点
- 自动检测新邮件
- AI智能内容分析
- 自定义分类规则
- 批量处理能力
- 详细的处理日志
技术分析
节点类型及作用
- @N8N/N8N Nodes Langchain.Lmchatopenai
- Gmailtool
- Code
- Stickynote
- Httprequest
复杂度评估
配置难度:
维护难度:
扩展性:
实施指南
前置条件
- 有效的Gmail账户
- n8n平台访问权限
- Google API凭证
- AI分类服务订阅
配置步骤
- 在n8n中导入工作流JSON文件
- 配置Gmail节点的认证信息
- 设置AI分类器的API密钥
- 自定义分类规则和标签映射
- 测试工作流执行
- 配置定时触发器(可选)
关键参数
| 参数名称 | 默认值 | 说明 |
|---|---|---|
| maxEmails | 50 | 单次处理的最大邮件数量 |
| confidenceThreshold | 0.8 | 分类置信度阈值 |
| autoLabel | true | 是否自动添加标签 |
最佳实践
优化建议
- 定期更新AI分类模型以提高准确性
- 根据邮件量调整处理批次大小
- 设置合理的分类置信度阈值
- 定期清理过期的分类规则
安全注意事项
- 妥善保管API密钥和认证信息
- 限制工作流的访问权限
- 定期审查处理日志
- 启用双因素认证保护Gmail账户
性能优化
- 使用增量处理减少重复工作
- 缓存频繁访问的数据
- 并行处理多个邮件分类任务
- 监控系统资源使用情况
故障排除
常见问题
邮件未被正确分类
检查AI分类器的置信度阈值设置,适当降低阈值或更新训练数据。
Gmail认证失败
确认Google API凭证有效且具有正确的权限范围,重新进行OAuth授权。
调试技巧
- 启用详细日志记录查看每个步骤的执行情况
- 使用测试邮件验证分类逻辑
- 检查网络连接和API服务状态
- 逐步执行工作流定位问题节点
错误处理
工作流包含以下错误处理机制:
- 网络超时自动重试(最多3次)
- API错误记录和告警
- 处理失败邮件的隔离机制
- 异常情况下的回滚操作