OpenAI - Firecrawl Docs

Integrate Firecrawl with OpenAI to build AI applications powered by web data.

Setup#

npm install firecrawl openai zod

Create .env file:

FIRECRAWL_API_KEY=your_firecrawl_keyOPENAI_API_KEY=your_openai_key

Note: If using Node < 20, install dotenv and add import 'dotenv/config' to your code.

Scrape + Summarize#

This example demonstrates a simple workflow: scrape a website and summarize the content using an OpenAI model.

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 });// Scrape the website contentconst scrapeResult = await firecrawl.scrape('https://firecrawl.dev', {    formats: ['markdown']});console.log('Scraped content length:', scrapeResult.markdown?.length);// Summarize with OpenAI modelconst 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);

Function Calling#

This example shows how to use OpenAI's function calling feature to let the model decide when to scrape websites based on user requests.

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'] // Other formats: html, links, etc.            });            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);}

Structured Data Extraction#

This example demonstrates how to use OpenAI models with structured outputs to extract specific data from scraped content.

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('Scraped content length:', 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: 'Extract company information from website content.'        },        {            role: 'user',            content: `Extract data: ${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('Validated company info:', companyInfo);

Search + Analyze#

This example combines Firecrawl's search capabilities with OpenAI model analysis to find and summarize information from multiple sources.

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 });// Search for relevant informationconst searchResult = await firecrawl.search('Next.js 16 new features', {    limit: 3,    sources: [{ type: 'web' }], // Other sources: { type: 'news' }, { type: 'images' }    scrapeOptions: { formats: ['markdown'] }});console.log('Search results:', searchResult.web?.length, 'pages found');// Analyze and summarize the key featuresconst 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);

Responses API with MCP#

This example shows how to use OpenAI's Responses API with Firecrawl configured as an MCP (Model Context Protocol) server.

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));

For more examples, check the OpenAI documentation.