AgentReader

Extract the full text of any PDF from a URL - no upload, no OCR setup, just pass the link. Returns plain text plus page count and word count as JSON. Built for agents reading research papers, filings, annual reports, datasheets and documentation. Handles files up to 25 MB.

Convert raw HTML you already have into clean markdown. Send JSON {"html": "..."} and get back markdown with scripts, styles and boilerplate stripped. Cheaper than /read when the agent has already fetched the page.

DNS lookup over Cloudflare DoH: query A, AAAA, CNAME, MX, NS, TXT, SOA, SRV, CAA or PTR records for any domain and get structured JSON answers with TTLs. Useful for domain research, email validation and infrastructure checks.

AI structured-data extraction: turn any web page, text or HTML into JSON matching YOUR schema. Send JSON with one of url/text/html plus 'schema' (a JSON Schema object) and optional 'instructions'. Returns validated JSON. Powered by Claude with guaranteed schema conformance.

Parse any RSS, Atom or RDF feed into normalized JSON: feed title, description and up to 100 items with title, link, publish date, author, summary and categories. Ideal for agents monitoring news sources, blogs and changelogs.

Extract every hyperlink on a web page as structured JSON: resolved absolute URL, anchor text, internal-vs-external classification and nofollow flag, up to 2000 unique links. The building block for agent crawlers that need to decide where to go next, plus broken-link sweeps, backlink and SEO analysis, and site mapping.

Get structured metadata for any public web page in one call: title, meta description, canonical URL, Open Graph and Twitter card tags, favicons, JSON-LD schema types, language and the heading outline. Use it to build link previews and social cards, classify or dedupe URLs, and run SEO and AI-readability audits - far cheaper than fetching and parsing the whole page.

Scrape any public web page and get back the main article as clean, LLM-ready markdown. Fetches the URL, strips nav, ads, cookie banners and footers with Readability, and returns JSON: title, byline, site name, publish date, excerpt, language, word count and the markdown body. The cheap way to feed real page content to an agent without burning tokens on raw HTML. Static-HTML extraction (no JS rendering) - use /screenshot when you need a rendered browser view.

OFAC sanctions screening: fuzzy-match a person/entity/vessel name (transliteration-aware - Arabic and Slavic romanization variants) or look up a crypto address against the official OFAC SDN and Consolidated lists. Returns verdict (match/possible_match/clear), scored matches with programs and aliases, and full list-version provenance. Lists re-ingested every 6 hours.

Parse an XML sitemap or sitemap index into JSON: up to 5000 URLs with lastmod and priority, or the list of child sitemaps. Useful for agents planning site crawls or auditing site structure.

AI summarization of any web page, raw text or HTML. Send JSON with one of url/text/html plus optional style (bullets|paragraph|tldr) and maxWords. Returns a dense, factual summary that preserves key numbers, names and conclusions. Powered by Claude.

Render any web page in a real headless browser and return a PNG screenshot. Supports viewport width/height (320-1920) and full-page capture. Ideal for agents doing visual verification, UI monitoring, archiving or content review.

Verified-delivery trust data for any x402 service: full history of real PAID test purchases against the endpoint (not free 402 probes) - delivery rate, response validity, latency, price stability. Free aggregate at /leaderboard; this endpoint returns the complete per-check history for one resource. Untracked resources return 404 and you are NOT charged. Use it before wiring an unknown x402 service into your agent.