Search & web

Answer a question when you have no URL yet: searches Wikipedia (source 'web') or recent news articles (source 'news') and returns the top 1 to 5 result pages already read as clean Markdown, with title, URL and snippet. Replaces one search call plus several page reads. Not a general web search engine. If links and a short summary are enough, web-search is cheaper. Charged only if at least one page was read.

Website crawler: read a whole website or one section of it (for example every page under /docs) as clean Markdown: starts at a URL, follows same-site links breadth-first and returns up to 20 pages with title and description. Use for documentation, product catalogs or company sites. For one page use web-scrape, for a known list of URLs web-read-batch. Charged only if at least one page was read.

List the URLs of a website without reading them: robots.txt, sitemaps and sitemap indexes plus homepage links, up to 1,000 same-site URLs with lastmod and source. A cheap first step to decide which pages to read with web-read-batch or site-crawl. Returns no page content. Params: url (site or domain, required), max_urls (1 to 1000, default 200).

Parse a User-Agent string into browser, engine, OS, device type/vendor/model and CPU, and flag bots and AI crawlers (GPTBot, ClaudeBot, PerplexityBot ...). Params: user_agent (required).

Read 2 to 5 pages whose URLs you already know in one call, each as clean Markdown with title and description, fetched in parallel. Cheaper and faster than calling web-scrape once per URL. A failed page is reported with its error while the others still return; charged only if at least one page was read. To find URLs first, use site-map (one site) or search-read (a question). Params: urls (1 to 5, required), max_chars per page (500 to 30000, default 6000).

Check whether a URL is reachable and what it returns: final status code, full redirect chain, response time, content-type, server header and content length. Use for link validation, uptime checks, canonical URL resolution or before spending money on a scrape/screenshot. Params: url (required), method (GET default, or HEAD).

Read ONE JavaScript-heavy page (single-page apps, dashboards, content loaded after page load) in a real Chromium browser and get the rendered text with title, description, word_count and optional links. Slower and pricier than web-scrape: use it only when web-scrape returns little or no content. Params: url (required), include_links, max_chars (default 8000).

Read any public URL as clean Markdown for an LLM: articles, docs, product pages, news, PDFs (text by page), RSS feeds and JSON. Real tables, code with language, math as TeX, absolute links; author, publish date and other metadata from meta tags and JSON-LD; optional links list; long pages in parts. Bot checks and empty pages are not charged. No account or API key. $0.002 USDC per call.

+4 more from Agent Utility Suite

Search planning applications anywhere in Ireland (all councils) by area name OR around a map point. Filter by outcome (granted/refused/pending), type, minimum housing units or a keyword like 'solar' or 'apartments'. Each result has a plain outcome field plus reference, council, address, works, dates and council link. Newest first.

What AI agents search for when they look for paid APIs and data: the last 30 days of searches on nohumans.directory, refreshed every 15 minutes, grouped by meaning. Each need has example phrasings, searches, distinct clients, and how many results they got (near zero = unmet demand). Use it to decide what paid API or dataset to build. Sellers checking their own listings and lookups of a service's own name are listed apart. A need shows only when 3+ clients asked.

URL to markdown: read any web page as clean, LLM-ready markdown. Use when an agent has a URL (article, docs, blog, product page) and needs the text without HTML, scripts or menus, e.g. to scrape article text or feed a page to an LLM. Pass url (https). Returns title, description, publish date, markdown, word count, headings and links. 0.005 USDC/call; failed calls are free.

Jev-powered research: routed search, cited answer, claims verified against sources. Ask a question, get a 5-8 sentence answer, each claim tagged to a numbered source, built from real page content - no invented facts. A fast paid LLM writes it, under 4.5s guaranteed; on a rare miss, a no-LLM extractive fallback (top sentences from the sources, picked by Jev) takes over instead, marked "synthesis":"extractive". Try GET /research/sample.

Fetch URL content as Markdown: read any web page and get clean Markdown - navigation, ads and boilerplate stripped, links resolved, plus a real token count. The same extraction /search uses on result pages, exposed standalone for a URL you already have. Try GET /web-read/sample. Part of the AgentIndex content kit (pdf, web-read, extract, summarize, detect-language) - see GET /capabilities.

Find up to five npm packages by query - version, description, ranking score and source links

Flesch Reading Ease and Flesch-Kincaid Grade Level scores for English text (standard public formulas, heuristic syllable counting).

Find academic papers in any field - title, authors, year, venue, DOI, citation count, open-access link and abstract - from OpenAlex (~250M works incl. arXiv and PubMed; Crossref answers if OpenAlex is down). Filter by year range and open access; sort by relevance, citations or date. Send {} to search the published example query. A search with no result is not charged. Pay-per-call via x402.

+2 more from FiatDock

Eyes for agents, for a penny: renders any URL in a real browser (Cloudflare Browser Rendering) and returns a full-page screenshot as an inline PNG plus an axe-core accessibility audit — WCAG violations with rule, impact, fix-it link, and affected-node counts. Verify a deploy visually, catch layout breakage, and check accessibility in one call, without running a browser yourself. https URLs only.

+1 more from MimirAPI

Search the scholarly literature by keyword or topic and get a ranked list of works with core metadata. Optional filters for year range, open-access only, concept and institution. Powered by OpenAlex. A cheap first step before /scholar/paper.

AI-search visibility (AEO) audit of a website for answer engines and agents. Fetches the target's own public surface and computes a composite visibility score 0-100 with a rating (poor, fair, good, excellent) across six axes: answerability, structured data, AI-crawler access, crawlability, llms.txt readiness and MCP well-known readiness, plus a prioritized list of fixes. Deterministic, no LLM. Visibility indicators, not a guarantee.

Search US motor carriers by name in the FMCSA Company Census and get ranked candidates, each with its USDOT number, MC docket, legal name, state, decoded carrier status, authority status and operation. Cheap first step before freight/carrier.

Structured parse of a public web page: main text, headings, links, tables and rich metadata (description, canonical, lang, Open Graph, Twitter cards, JSON-LD) as separate fields. Browserless, robots.txt-respecting.

Search foods by name in USDA FoodData Central (public domain). Returns ranked matches with fdc_id, description, brand, GTIN and key macros per 100g.

Stack Exchange question search with score, answer count and whether an answer was accepted

+9 more from tokenguard

Search and synthesize scientific literature on longevity and aging. Submit a research question; returns cited answer with references from PubMed and other sources.

+1 more from x402-api.aubr.ai

Live video search: ranked video results (title, video page URL, description, duration, creator, publisher, thumbnail, age) from an independent search index as clean JSON. Completes the web/news/images search family. Same freshness filter (pd/pw/pm/py); strict safe-search is on by default.

Pull a subset of pages into a new PDF (split). Body: {"url":"https://…/file.pdf","pages":"1-3,5"}. Returns the new PDF as base64.

The 'who just filed / who just bought / what just IPO'd / what does the full-text search say' monitoring loop. Five EDGAR tools composed: full-text search for a keyword, surface recent filings on the matched companies, pull insider Form 4 transactions, pull 13F institutional holdings changes, and check the IPO calendar. Different from sec-filings-deep-dive (which goes deep on one company's earnings) - this is the wide-and-recent radar.

What companies are going public? Recent S-1/IPO filings from EDGAR plus a web search for IPO news. IPO watch skill pack: one x402 payment runs 3 underlying tools (edgar-recent-ipos, search, search-news); partial-success per step.

Pull structured data out of any web page deterministically - articles to clean text, tables to JSON rows, specific elements via CSS selector - without writing regex against raw HTML. Structured scrape skill pack: one x402 payment runs 7 underlying tools (extract, render, html-select, html-table, html-strip, html-links, html-meta); partial-success per step.

Web search that returns READY-TO-USE grounding text, not links: for one query you get the passages an independent search index extracted from the pages it ranked, flattened into chunks with their source URL and title and an approximate token count, under a token budget you choose (1024-32768, default 4096). Use it to ground a model or a RAG pipeline in one call, where search + extract would take several.