Agent Extraction API

Base or Ethereum contract address -> ABI and function list: Sourcify verification (exact/partial match), contract name, compiler, ABI, function selectors + signatures, event topics; proxies (EIP-1967, ZeppelinOS, others Sourcify detects) resolved to the implementation with its own ABI. Unverified contracts: selectors guessed from bytecode and named via the public OpenChain database. abi=false returns signatures only (smaller).

Country reference lookup by ISO2/ISO3/name: capital, population, area, continent, currency, phone prefix, TLD, languages, neighbours, time zones with current UTC offset and local time.

Currency conversion with official ECB euro reference rates (~30 currencies, daily). Params: from, to, amount. Also returns the rate date.

DNS lookup over DNS-over-HTTPS (Cloudflare, Google fallback): A, AAAA, MX, NS, TXT, CNAME, CAA, SOA, SRV, HTTPS, DS... for one domain in one call, plus a mail-auth check (MX hosts, SPF, DMARC policy, warnings) and the DNSSEC-validated flag.

Domain registration lookup via RDAP (the official whois successor, IANA bootstrap): registered or not, registrar + IANA id + abuse email, created / updated / expiry dates, days to expiry, domain age, EPP status codes, nameservers, DNSSEC. Use it to vet a domain or check expiry.

ENS + Basenames resolver: name -> address (vitalik.eth, jesse.base.eth) or address -> verified primary name on Ethereum and Base (reverse record checked forward), plus optional text records (avatar, url, com.twitter, com.github, ...). On-chain via public RPC, no keys.

RSS / Atom / JSON Feed to clean JSON: give a feed URL (or a web page that links one, auto-discovered) and get feed title/link/language plus normalised items: id, title, link, ISO dates, author, plain-text summary, categories, enclosures (podcast audio, media), image. Filter with since= and limit=; include_content=true adds the full text.

Geocoding with OpenStreetMap Nominatim: address or place name -> coordinates (up to 5 matches, optional country filter), or lat+lon -> structured address (reverse). Structured address fields, country code, bounding box, OSM id. Cached; ODbL attribution included.

Public holidays for ~490 countries/territories for 2026 and 2027; optional date=YYYY-MM-DD answers is_holiday.

Invoice or receipt (PDF, PNG/JPEG/WebP image, or plain text) to structured JSON with a fixed schema: vendor (name, address, VAT ID, email, phone, IBAN), customer, invoice number, order number, issue/due date, currency (ISO 4217), subtotal, VAT total, total, amount due, VAT breakdown and line items. Every field has value, the verbatim source quote and a confidence score (0-1: quote found in the document, value matches quote, totals add up). Arithmetic checks included. $0.0075 per page (max 10 pages per call; the 402 quotes pages x $0.0075), 1 free call per caller (up to 3 pages). Runs on Cloudflare Workers AI only (PDF text layer via toMarkdown, OCR for images); documents are not stored. Scanned PDFs without a text layer: send page images. POST JSON {url} or {file_base64, mime_type} or {text}, or POST the raw file (Content-Type application/pdf or image/*); GET ?url= or ?text= also works.

Check up to 20 URLs in one call: HTTP status, final URL after redirects, redirect count, latency, content-type, ok flag.

Token & cost calculator: estimate tokens for a text (or pass input_tokens/output_tokens) and price it across several models using provider list prices (LiteLLM price table, MIT). Params: models (comma list of model ids like gpt-4o-mini, claude-sonnet-4-5, gemini/gemini-2.5-flash or search words, max 15), text or input_tokens, output_tokens, calls.

One-call page audit for agents: HTTP status, redirects, latency, title, meta description, canonical, lang, Open Graph, JSON-LD types, headings, word count, plus agent-friendliness: robots.txt rules for 14 AI crawlers, llms.txt, sitemap, and a 0-100 score with reasons.

Fetch a public web page (robots.txt respected) and return title, meta description, canonical, Open Graph/Twitter tags, language, headings, link counts, HTTP status.

Any public web page -> clean LLM-ready Markdown + RAG chunks in one call: main content via Mozilla Readability (no nav/ads/footers), absolute links, title/description/language/canonical/published+modified dates/JSON-LD types, word + token estimates, sha256 content hash for change detection, and heading-aware chunks (configurable size + overlap, each with its heading path). Plain HTTP, robots.txt respected, Markdown/plain text passed through. Flags pages that likely need JavaScript.

QR code generator: text, URL, payment URI (e.g. ethereum:/USDC request) or Wi-Fi credentials (ssid + password) to a QR code as SVG (inline + data URI) and/or GIF data URI. Error correction L/M/Q/H, scale, quiet-zone margin, colours, optional 0/1 module matrix. UTF-8 byte mode, computed locally, nothing stored.

Check whether a URL may be crawled per the site's robots.txt for a given user-agent. Returns allowed, matched rule, crawl-delay, sitemaps.

Company website -> one clean contact row: role emails only (info@, sales@ ...) with MX check, phones validated and normalized to E.164 with confidence, schema.org address/name, Danish CVR and EU VAT ids, official social profiles. Respects robots.txt, plain HTTP, home + up to 5 contact pages. Personal emails omitted (GDPR).

Sitemap to URL list: give a sitemap URL or just a domain (sitemaps found via robots.txt, then /sitemap.xml). Follows sitemap indexes (newest first, up to 10 child sitemaps), reads .xml, .xml.gz and text sitemaps, and returns every URL with lastmod, changefreq, priority, hreflang alternates, image and news counts. Filter by contains= and since= (lastmod).

TLS/SSL certificate check from public Certificate Transparency logs (Cert Spotter, crt.sh fallback): currently valid certificates for a hostname, newest first, with issuer, validity dates, days left, SAN names and an expiring-within-14-days flag. For expiry monitoring and phishing/subdomain vetting.

ERC-20 token lookup on Ethereum, Base, Arbitrum, Optimism or Polygon: name, symbol, decimals, total supply (raw + human), live USD price with confidence (DefiLlama) and FDV, for up to 10 token addresses in one call. Use it before showing, pricing or converting a token amount.

Deterministic ERC-20 risk check on Base or Ethereum from public on-chain data only: privileged functions in bytecode (mint, blacklist, pause, fees, limits, trading switch), upgradeable proxy + implementation, owner and renounce status, Uniswap v2/v3 + Aerodrome liquidity in USD (Chainlink), owner/pool/burned supply share. Every flag has evidence; lists what was NOT checked (no honeypot simulation). Not financial advice.

Base or Ethereum transaction hash -> plain JSON: status (success/failed/pending), block time, confirmations, from/to, ETH value, gas used and fee (incl. Base L1 fee), method selector + name, and every ERC-20/ERC-721 Transfer in the receipt with token symbol, decimals and human amounts. Optional payment check: expect_to + expect_token (USDC, ETH, symbol or address) + expect_amount confirms a payment arrived (e.g. verify a USDC or x402 settlement). Public RPC data only.

Transaction status for polling: up to 10 tx hashes in one call, searched across Ethereum, Base, Arbitrum, Optimism and Polygon (or one chain): pending / success / failed / not_found, block, confirmations, finalized flag, from/to, gas used, fee in native coin, explorer link.

Base or Ethereum address -> profile in one call: EOA, contract or EIP-7702-delegated EOA; ETH balance; nonce (tx count); balances of major tokens (USDC, USDT, EURC, DAI, WETH, cbBTC/WBTC, USDbC); for contracts: bytecode size, EIP-1967/ZeppelinOS proxy implementation, beacon, owner(), ERC-20 name/symbol/decimals/supply and Sourcify verification. Public RPC data only.