Context.dev design system

Crawl a site in three steps.

Point it at a domain, choose the formats, and read every page back as clean Markdown.

1Get an API keySign up and copy your key from the dashboard.
2Start a crawlSend the start URL and how many pages to follow.
3Read the pagesEach page comes back as Markdown, HTML or JSON.

$ npm install context.dev

Free to start: 1,000 credits a month, no card needed.

crawl.ts

import ContextDev from 'context.dev';const client = new ContextDev({ apiKey: process.env.CONTEXT_DEV_API_KEY });const crawl = await client.web.crawl({  url: 'https://docs.stripe.com',  maxPages: 500,  formats: { markdown: true },});for (const page of crawl.pages) {  console.log(page.url, page.markdown.length);}

from context_dev import ContextDevclient = ContextDev(api_key=os.environ["CONTEXT_DEV_API_KEY"])crawl = client.web.crawl(    url="https://docs.stripe.com",    max_pages=500,    formats={"markdown": True},)for page in crawl.pages:    print(page.url, len(page.markdown))

curl https://api.context.dev/v1/web/crawl \  -H "Authorization: Bearer $CONTEXT_DEV_API_KEY" \  -d '{ "url": "https://docs.stripe.com", "maxPages": 500 }'

From URL to answer.

Search

Find the pages that answer the question.

const results = await client.web.search({  query: 'stripe usage based billing',});

Scrape

Read each result as Markdown.

const page = await client.web.scrape({  url: results[0].url,  formats: { markdown: true },});

Answer

Hand the pages to your model with their sources.

const answer = await model.generate({  context: page.markdown.data,});