Search & web

Bulk Bitcoin block extract (max 144 blocks per call, flat price per call). Returns array of per-block summaries (height, time, miner pool, tx count, witness/taproot/ordinals/op_return stats, total output, coinbase reward).

+2 more from api.carbon-cashmere.de

Web search plus page reading in one call: answer with sources. Runs a web search, scrapes the top results, and returns the ranked search results together with each page's content as clean LLM-ready markdown, so an agent gets the SERP AND the underlying text without a second round of scrape calls. Per-page fetch errors are reported honestly per URL.

Scrape and extract clean text/markdown from a public URL for agent pipelines on Venice, paid per request in USDC on Base via x402, no API key. Standalone; no model lookup required.

+1 more from AiSpace Venice

Checks how well a Shopify store's products show up when AI shopping assistants such as ChatGPT, Gemini, Perplexity, Grok, and Copilot search for products, and reports the store's rank and its closest competitors. Searches Shopify's real Global Catalog - the same product index AI assistants read from.

Fetch deterministic, token-efficient context from one public HTTPS URL: canonical URL, title/description, headings, clean main content, structured-data types, freshness observations, evidence, and limitations. Call when you need a bounded JSON reading of one public page without a browser, login, crawl, or LLM.

+2 more from Agent Mercantile

Give this API a link to any website's sitemap.xml file and it will fetch and parse it, returning every URL found inside. If the sitemap links to other sitemaps (called sitemap indexes), it will follow and extract those too. Each URL is returned with any extra information available, such as when the page was last updated and how often it changes. Useful for SEO audits, content inventories, broken link checks, and competitive research.

Give this API a web address and it will visit that page, then hand back everything an SEO professional needs to know: the page title, meta description, meta keywords, robots instructions, canonical link, H1/H2/H3 headings, how many links are on the page, and the HTTP status code. No setup required — just paste a URL and get structured data back in under two seconds.

Extract a public web page into timestamped, normalized evidence.

Search consumer-product safety notices by authority, jurisdiction, recall status, category, risk, hazard, manufacturer, date, or free text.

Choose a context budget before an agent run. Compare five extractive compilation budgets with token estimates and source coverage. Price basis: five compiler runs, the comparison table and the receipt digest. Uses caller-supplied documents; no answer generation, factual verification or guaranteed token-billing savings.

Extract up to ten URLs in one paid call.

Fetch a public HTTP(S) page with private-network blocking and extract its title, description, Open Graph image, status, and content type.

Canonicalize URLs: strip tracking params, force HTTPS, sort query, extract components.

+6 more from Grokzilla x402 Micro Pay

Web page to Markdown: fetch a public URL and get clean Markdown plus structured JSON (title, meta description, language, headings, tables, links, Open Graph, JSON-LD). Deterministic, no AI, fast. Respects robots.txt; you are only charged if the page was fetched successfully. (Webseite → Markdown + strukturierte Daten.)

Keyword extraction and named entity recognition (NER) from text: ranked keywords plus people, organizations, locations, dates, money and products. Any language, clean JSON output.

Extract structured JSON from unstructured text (invoices, emails, ads, CVs...): name the fields you want, get clean JSON back. Any language, up to 6,000 characters.

Supporting helper: SSRF-safe public URL → title, excerpt, links. Secondary to Daily Agent Signal.

SEC EDGAR Full-Text Search ($0.02): search the actual TEXT of SEC filings since 2001, across any company, by phrase — not just per-company metadata. Fetched live from SEC's own full-text search index (efts.sec.gov), not a snapshot. Body { q, forms?, ticker?, since?, until?, limit?: 1..40 }. Check free /v1/edgar_fulltext_search/availability first.

Web search API (SERP) with page text: like a search API key, pay per call. Pass q, optional n (1-10, default 5). Returns ranked results with title, url, domain, snippet, publish date and up to 1500 chars of cleaned main text from each page, fetched live, so agents can answer without opening links. Fresh results for news, docs, products, local businesses, definitions. 0.01 USDC/call; failed calls are free.

Extract structured fields (email, phone, dates, amounts, invoice numbers, …) from any text via deterministic regex — no LLM, no API key. $0.02 USDC per call on Base.

Search Aggregator deep tier: Serper+Tavily+Exa+Firecrawl merged results with full page content and synthesized answer, fixed price $0.05 USDC. POST JSON body {query (required), max_results (optional 1-20, default 10)}. Mode is fixed by the path — no mode field needed, the quoted price is always the settled price.

+1 more from Search Aggregator

Read a public static HTTPS webpage as clean text with headings, links, timestamp and content hash. Use to ground an answer or ingest documentation. Respects robots.txt; no JavaScript or login.

+3 more from Evidence402

Extract PDF/CSV/XLS/XLSX/ZIP/JSON/XML links from static public HTML by suffix or download filename. Resolve relative/base links and deduplicate. Limits: 500 links, 8s, 5 redirects, 2 MiB. No JavaScript or file downloads; existence, MIME, size and file safety are unverified. Empty results are valid. 0.003 Base Sepolia test USDC per success; invalid inputs/errors are not settled.

+2 more from Forest Gas Station

Compile 1-8 documents into a token-budgeted context pack, then measure exact retention of 1-32 caller-labelled evidence spans. Returns the pack, source-linked passages, input and output hashes, original and compiled token estimates, tokensSaved, source coverage, and requiredEvidenceRetentionPercent. Retention is exact span matching only: not accuracy, answer quality, verification, or hallucination prevention. Extractive and budget-bound. Request content is not stored.

Search scientific papers across OpenAlex (250M works, all fields), arXiv (preprints) and PubMed (biomedicine) in one call. Returns title, up to 5 authors, year, venue, DOI, landing URL, open-access PDF URL, citation count and abstract (up to 600 chars), merged and deduplicated by DOI. Params: query (required), source ('all' default, 'openalex', 'arxiv', 'pubmed'), limit (1 to 25, default 10), year_from, open_access_only (default false).

Extract one bounded value from caller-supplied JSON using a deterministic simple path such as items[2].name or $.data.id. Forest performs no network request or model inference, returns an explicit found flag, and supports only dot object keys plus numeric array indexes so lookup behavior stays predictable. Base mainnet USDC

Extract public SKU, GTIN, UPC, EAN, ISBN, MPN and provider identifiers from a product source. Missing identifiers stay null and are never guessed. Base mainnet USDC

SEO title tag and meta description from a URL: reads the page and returns the current title and meta description with their lengths, a better title (max 60 characters), a meta description of 140 to 155 characters with a reason to click, an H1, 5 to 8 search phrases and the problems found. Optional keyword to build around (e.g. "windows Munich") and language. Written by Claude Haiku 5.5 from the page content only, no claims the page does not make. Not charged when the page cannot be read.

Extract structured JSON from a web page or text with fields you define: send url (or text) and fields like {"price": "price in EUR as number", "in_stock": {"type": "boolean"}}, get back exactly those fields, typed and validated against a JSON schema. Types: string, number, integer, boolean, string[], number[]. Fields the content does not state come back as null instead of being guessed. Up to 25 fields; optional instructions for extra rules.

Summarize any web page: send a URL, get a factual summary that starts with what the page is, followed by the key facts, numbers and conclusions as bullet points. Reads the page (HTML to Markdown) and summarizes it with Claude Haiku 5.5. Options: length short (60 to 90 words), medium (150 to 220) or long (350 to 450); language for the summary (default: the page language, e.g. German); focus to steer it (e.g. "pricing and plans").