web-search-and-content

Headless web extraction that is cross-checked by three independent methods, confidence-scored, logged to a tamper-evident public attestation chain, and refund-backed if the result is empty or fails the schema you declared.

Extract structured data from content you already have (e.g. from a prior /content call), using AI. No browser render involved -- pure AI extraction, cheaper and faster than fetching a URL fresh. Use this when you already have the text; use /extract-url if you only have a URL.

Fetch a URL with a real rendered browser and extract structured data from it in one call, using AI (tries multiple models in sequence for reliability). Use this when you don't already have the page content; use the cheaper /extract if you do. Results are cached for 24 hours per URL+schema combination.

Web extraction: every output you ask for in `formats`, in one call.

Extract text from an image by URL (OCR): PNG/JPG/WEBP/TIFF/BMP of a screenshot, scan, receipt, chart or photo -> the readable text, via the Tesseract engine. Keyless, no LLM. Use it to read text an agent can only see as an image. SSRF-guarded; max ~10MB.

Fetch any public URL and return a clean, readable markdown or JSON extract for AI agents. Pass the target as ?url=<encoded>. Ideal when you need to read webpages, docs, or articles without a browser. Blocks script/style tags, normalizes links and images, strips HTML.

Full-text search across the entire indexed public-domain corpus. Returns the most relevant passages from any book with title/author citations, ranked by relevance. The premium cross-corpus retrieval route — no LLM generation, real excerpts with sources. No license, no API key required.

Web page to clean text: title, description, headings, links and readable text from any public URL — Genesis402 / UnyKorn Operator Network

URL -> AI-readiness report. Checks schema.org structured data, crawlability (robots.txt, meta robots, canonical), and llms.txt presence, then returns a 0-100 score with specific recommendations for what's missing. Results are cached for 24 hours per URL.

AI-visibility check: which AI answer engines cite a domain for a prompt, via their official APIs.

Fetch any URL and get clean, LLM-ready page content — no API key, no account, no signup; pay per call in USDC on Base. Gets past common bot-blocking. Params: ?url=... (required), format=markdown|text|html (default markdown), max_chars=N (default 40000, max 120000). Returns { url, status, title, description, format, content, truncated, chars, latency_ms }; non-HTML URLs return raw content. For agents that need a page as text before summarising or extracting.

Search: live web search results as clean JSON.

Deep research pack: 3 searches + up to 10 pages as Markdown (Research-пакет deep: 3 поиска и до 10 страниц в Markdown)

Keyless web search for AI agents. Give it a natural-language query and it returns ranked organic results (title, url, snippet) drawn from live multi-source search (DuckDuckGo HTML/Lite plus public SearXNG instances, with automatic fallback so it never returns empty when one source flakes). Optionally it also fetches the top result and extracts the main content as clean, token-budgeted markdown (answer_extract), so an agent can read the page without running its own crawler, browser or scraper. Use it to answer real-world questions, gather fresh facts, research a topic, find documentation or check what a site says, without maintaining a search index or paying for Google/Bing/Serper API keys. Resilient (short timeouts, multi-source), SSRF-hardened, 0.005 USDC on Base.

Screenshot API for AI agents: send {url} and get a JPEG, PNG, WebP or PDF of the page as base64 in JSON, with final URL, byte size and pixel size. Pick a desktop, mobile or tablet viewport, full page, dark mode, a delay or one element by CSS selector. It shows what a normal browser sees and does not get past bot protection (use unblock for those pages' text). A page that can't be captured, or renders blank, is not charged.

Fetches one URL and returns the readable page as markdown: title, description, headings, paragraphs and links with their destinations, plus the canonical URL, language and word count. Boilerplate, scripts and navigation are stripped. Built for going through a list of links one at a time. No API key, no account.

Structured extraction of one web page: title, description, headings, clean main text, markdown, links and images.

AI-package release velocity. Per-package releases_24h / releases_7d / releases_30d, latest bump classification (major/minor/patch/prerelease, pre-1.0 minor counts as major per semver), is_breaking_recent flag (major bump within 30 days). Notable movers: recent_major_bumps, most_releases_7d, fastest_cadence_30d. Filter by ecosystem (PyPI/npm), category (llm-sdk, agent-framework, etc), or package substring. The "what AI deps just changed and is it breaking" call.

Structured web extract for AI agents: fetch any public URL, return schema-constrained JSON for SaaS pricing, contacts, team, product, FAQ, or task-shaped fields. Confidence scores and source snippets. No API key — pay-per-call x402 USDC on Base ($0.02). Better than raw crawl markdown for agent tools.

Transcribes an audio file with Whisper large v3 turbo: pass an https URL (mp3, wav, flac, m4a, ogg, up to 10 MB) or base64 audio in a POST. Returns the text, the detected language with its probability, the duration, and timed segments; add words=true for word-level timestamps. Around 100 languages. You are only charged if the transcript is delivered. No API key, no account.

Scrape any URL to clean markdown (Firecrawl-backed). POST { url } → { url, title, markdown }.

Google web search (Serper-backed). POST { query, num? } → { results: [{title,url,snippet}] }. No API key, no subscription — pay per call.

Semantic/neural web search (Exa-backed). POST { query, num? } → { results: [{title,url,snippet}] }.

Read any public web page as clean Markdown: main-content extraction (navigation, footers, scripts and hidden elements removed), headings, lists, tables, code blocks, links (absolute, deduplicated), images, plus title, description, canonical URL, language, author and publish date, and a content hash + fetch time as evidence. JSON and plain-text URLs are returned as-is. No JavaScript rendering. $0.001 per page.