From a URL to facts your agent can use.
Read, map and research any company's site, and hand the model data in the shape its next step expects.
Markdown, not markup
Pages arrive without navigation, footers or cookie bars, so tokens go to the content.
JSON in your own schema
Describe the fields with a prompt or a JSON Schema, and anything absent comes back null.
Tools over MCP
Claude, ChatGPT and Codex call the same endpoints as tools, through one MCP server.
task
Write an account brief for stripe.com.
- prefetch_mapstripe.com search=aboutlinks
- prefetch_scrapestripe.com/aboutmarkdown
- prefetch_scrapeformats=jsonjson
context for the next step
{ "company": "Stripe",
"founded_year": 2010,
"headquarters": "South San Francisco",
"products": ["Payments", "Billing", "Connect"] }
How it works
Find the pages
Map a site's URLs, filtered by a search term, without fetching a single page.
GET /mapRead them
Scrape a page, or crawl a whole section, into clean markdown with every link made absolute.
GET /scrapeExtract the fields
Ask for the json format with a prompt or a schema, and get typed data back instead of prose.
formats=jsonFill the gaps
For facts no single page holds, /answers searches the web and returns sourced JSON.
POST /answers
Give agents context with Claude
Let an agent call Prefetch as a tool, or ask for JSON in your own shape with your API key.
# claude.ai/directory/prefetch → Connect
# Sign in to Prefetch, Allow access. No key to paste.
# Claude Code on the same Claude account has it too.
# Then just ask:
Use prefetch_scrape on linear.app/pricing with the
json format and give me each plan's name, monthly
price and seat limit.The endpoints behind it
One key and one pool of credits for all of them.
- Web Scraping APIAny page as markdown for the context window, or as JSON in your schema.GET /scrape
- Sitemap APIEvery URL a site exposes, so the agent reads the right pages and no others.GET /map
- Web Crawling APIA docs site or a whole section, rendered page by page into markdown.POST /crawl
- Web Research APIA research question in, a sourced answer in your JSON shape out.POST /answers
- Website Monitoring APIPages re-checked on a schedule, with a webhook carrying only what changed.POST /monitor
1 credit per page
Summaries and JSON add 3. Failed requests are free.
Free
USD
/month
1,000 credits / month
- Up to 1,000 pages
- 1 concurrent request
- 30 requests per minute
- No credit card required
Starter
USD
/month
10,000 credits / month
- Up to 10,000 pages
- 10 concurrent requests
- Unlimited requests per minute
- Email support
- $5 per 2,500 additional credits
The same plan unlocks every other endpoint too. Each line is the whole month of 10,000 credits spent there instead of on pages. Mix them however you like.
10,000 name lookups: Company Name to Domain API
(1 credit per lookup, a company, brand or product name in, its official website's domain out)
10,000 site maps: Sitemap API
(1 credit per call, every URL a site exposes, from its sitemaps and links, without fetching a page)
10,000 crawled pages: Web Crawling API
(1 credit per page, a whole site, or one section of it, rendered page by page into clean markdown)
10,000 screenshots: Screenshot API
(1 credit per call, any URL to a hosted image, at any viewport, full page or above the fold)
10,000 page checks: Website Monitoring API
(1 credit per page checked, pages re-checked on your schedule, with a webhook carrying only what changed)
2,000 company lookups: Company Data API
(5 credits per call, the legal entity, its size, founding year and published contacts behind a domain)
2,000 classifications: Website Classification API
(5 credits per call, the IAB category, subcategory and a readable description of any URL)
1,000 brand lookups: Brand API
(10 credits per call, logos, colors, fonts and social assets pulled from any domain)
1,000 answers: Web Research API
(10 credits per answer, a research question in, a sourced answer in your own JSON shape out)
833 enrichments: Company Enrichment API
(12 credits per call, brand, company and classification for any domain, from one page load)
Failed requests are never charged. Top-up credits are spent after your monthly credits and never expire.
AI Apps & Agents questions
Scale with confidence. Handle thousands of requests per second on fast, reliable infrastructure.
- Registered Users
- 300+
- API calls processed
- 40k
- Uptime
- 99.9%
Ready to get started?
Create an account instantly, or contact us to design a custom package for your business.
Up in 5 minutes
Make your first API call in under 5 minutes. No setup headaches, just copy, paste, ship.
API reference
Request schemas, response examples, everything you need to integrate.