将 Firecrawl 与 OpenAI 集成，构建由网页数据驱动的 AI 应用。
安装与配置#
npm install firecrawl openai zod创建 .env 文件：
FIRECRAWL_API_KEY=your_firecrawl_keyOPENAI_API_KEY=your_openai_key注意： 如果使用 Node 版本低于 20，请安装
dotenv，并在代码中添加import 'dotenv/config'。
抓取 + 摘要#
此示例演示一个简单流程：抓取网站，并使用 OpenAI 模型对内容进行摘要。
import { Firecrawl } from 'firecrawl';import OpenAI from 'openai';const firecrawl = new Firecrawl({ apiKey: process.env.FIRECRAWL_API_KEY });const openai = new OpenAI({ apiKey: process.env.OPENAI_API_KEY });// 抓取网站内容const scrapeResult = await firecrawl.scrape('https://firecrawl.dev', { formats: ['markdown']});console.log('Scraped content length:', scrapeResult.markdown?.length);// 使用 OpenAI 模型生成摘要const completion = await openai.chat.completions.create({ model: 'gpt-5-nano', messages: [ { role: 'user', content: `Summarize: ${scrapeResult.markdown}` } ]});console.log('Summary:', completion.choices[0]?.message.content);函数调用#
此示例展示了如何使用 OpenAI 的函数调用功能，让模型根据用户请求自动判断何时抓取网站内容。
import { Firecrawl } from 'firecrawl';import OpenAI from 'openai';import { z } from 'zod';const firecrawl = new Firecrawl({ apiKey: process.env.FIRECRAWL_API_KEY });const openai = new OpenAI({ apiKey: process.env.OPENAI_API_KEY });const ScrapeArgsSchema = z.object({ url: z.string().describe('The URL of the website to scrape')});const tools = [{ type: 'function' as const, function: { name: 'scrape_website', description: 'Scrape content from any website URL', parameters: z.toJSONSchema(ScrapeArgsSchema) }}];const response = await openai.chat.completions.create({ model: 'gpt-5-nano', messages: [{ role: 'user', content: 'What is Firecrawl? Visit firecrawl.dev and tell me about it.' }], tools});const message = response.choices[0]?.message;if (message?.tool_calls && message.tool_calls.length > 0) { for (const toolCall of message.tool_calls) { if (toolCall.type === 'function') { console.log('Tool called:', toolCall.function.name); const args = ScrapeArgsSchema.parse(JSON.parse(toolCall.function.arguments)); const result = await firecrawl.scrape(args.url, { formats: ['markdown'] // 其他格式:html、links 等 }); console.log('Scraped content:', result.markdown?.substring(0, 200) + '...'); // Send the scraped content back to the model for final response const finalResponse = await openai.chat.completions.create({ model: 'gpt-5-nano', messages: [ { role: 'user', content: 'What is Firecrawl? Visit firecrawl.dev and tell me about it.' }, message, { role: 'tool', tool_call_id: toolCall.id, content: result.markdown || 'No content scraped' } ], tools }); console.log('Final response:', finalResponse.choices[0]?.message?.content); } }} else { console.log('Direct response:', message?.content);}结构化数据提取#
此示例演示如何使用支持结构化输出的 OpenAI 模型，从爬取内容中提取特定数据。
import { Firecrawl } from 'firecrawl';import OpenAI from 'openai';import { z } from 'zod';const firecrawl = new Firecrawl({ apiKey: process.env.FIRECRAWL_API_KEY });const openai = new OpenAI({ apiKey: process.env.OPENAI_API_KEY });const scrapeResult = await firecrawl.scrape('https://stripe.com', { formats: ['markdown']});console.log('抓取的内容长度:', scrapeResult.markdown?.length);const CompanyInfoSchema = z.object({ name: z.string(), industry: z.string(), description: z.string(), products: z.array(z.string())});const response = await openai.chat.completions.create({ model: 'gpt-5-nano', messages: [ { role: 'system', content: '从网站内容中提取公司信息。' }, { role: 'user', content: `提取数据: ${scrapeResult.markdown}` } ], response_format: { type: 'json_schema', json_schema: { name: 'company_info', schema: z.toJSONSchema(CompanyInfoSchema), strict: true } }});const content = response.choices[0]?.message?.content;const companyInfo = content ? CompanyInfoSchema.parse(JSON.parse(content)) : null;console.log('已验证的公司信息:', companyInfo);搜索 + 分析#
此示例将 Firecrawl 的搜索功能与 OpenAI 模型的分析能力相结合，用于从多个来源中查找并汇总信息。
import { Firecrawl } from 'firecrawl';import OpenAI from 'openai';const firecrawl = new Firecrawl({ apiKey: process.env.FIRECRAWL_API_KEY });const openai = new OpenAI({ apiKey: process.env.OPENAI_API_KEY });// 搜索相关信息const searchResult = await firecrawl.search('Next.js 16 new features', { limit: 3, sources: [{ type: 'web' }], // 其他来源:{ type: 'news' }、{ type: 'images' } scrapeOptions: { formats: ['markdown'] }});console.log('Search results:', searchResult.web?.length, 'pages found');// 分析并总结关键功能const analysis = await openai.chat.completions.create({ model: 'gpt-5-nano', messages: [{ role: 'user', content: `Summarize the key features: ${JSON.stringify(searchResult)}` }]});console.log('Analysis:', analysis.choices[0]?.message?.content);使用 MCP 的 Responses API#
本示例展示了如何在将 Firecrawl 配置为 MCP (模型上下文协议) 服务器后使用 OpenAI 的 Responses API。
import OpenAI from 'openai';const openai = new OpenAI({ apiKey: process.env.OPENAI_API_KEY });const response = await openai.responses.create({ model: 'gpt-5-nano', tools: [ { type: 'mcp', server_label: 'firecrawl', server_description: 'A web search and scraping MCP server to scrape and extract content from websites.', server_url: 'https://mcp.firecrawl.dev/v2/mcp', headers: { Authorization: `Bearer ${process.env.FIRECRAWL_API_KEY}`, }, require_approval: 'never' } ], input: 'Find out what the top stories on Hacker News are and the latest blog post on OpenAI and summarize them in a bullet point format'});console.log('Response:', JSON.stringify(response.output, null, 2));更多示例，请参阅 OpenAI 文档。