All projects on Base matching “extraction”

Extract raw artifacts from images: OCR text and detected objects. No interpretation, just clean machine-readable data. Powered by GPT-4o Vision. Payment via USDC on base.

Use when you need to check whether an IBAN is valid and extract its country, bank code and check digits.

URL-to-Markdown web extraction and context compression for AI agents. Fetch a public web page and return clean, task-relevant, citation-verifiable LLM context with exact excerpts, locators, hashes, and estimated tokens saved. Use before an agent summarizes, answers questions from, or cites a long or noisy URL. Deterministic and model-free. Public HTML only; no PDFs, logins, or JavaScript-only page

Fetch a public URL with SSRF protection, clean extraction, and semantic prompt-injection analysis.

Fill a PDF form using facts found in supporting documents. Also: fill a PDF form from DOCX, spreadsheets, images, JSON, or other documents; pre-fill a government form from supporting paperwork; complete an application from a resume or company details; extract answers from context documents into a form. Price scales with page count; this 402 carries the exact amount for the PDF sent.

Browser Use: AI browser automation agents in the cloud. POST a natural-language task and an agent navigates websites, fills forms, clicks buttons, extracts structured data, and completes multi-step workflows just like a human. Stealth-enabled remote browsers with CAPTCHA solving, residential proxies, live session view, and optional structured output. Payments top up your balance; managing your ses

Fetch a URL and extract structured metadata from its HTML: title, description, canonical URL, language, favicon URL, RSS/Atom feed links, Open Graph tags (og:title, og:image, og:type, og:description, og:url), Twitter Card tags, and other meta tags (author, keywords, robots). Use this to get Open Graph data, preview cards, or site information without parsing HTML yourself. Send {"url": "https://...

Extract a readable article from supplied HTML, including title, author, language, metadata, clean Markdown, links and tables. Uses Mozilla Readability with a fallback for short pages. No remote fetches or script execution; URLs resolve relative links.

New @ai-mail.sh email inbox for an AI agent: 24 hours, up to 10 incoming emails. Receive-only; codes and verification links extracted. Docs: https://ai-mail.sh/llms.txt

Email inbox for an AI agent to receive signup verification codes and links (extracted for you). 24h, $0.02, up to 50 emails. Returns address + token; GET /v1/inboxes/{id}/code returns the code. Receive-only.

Lightweight utility API for agents and operators: DNS lookups, RDAP/WHOIS domain data, HTTP probes, page metadata extraction, cryptographic hashing, JSON validation, scheduled webhooks, and exotic endpoints (Base transaction decode, ENS resolve, USGS earthquakes, cron next-run, TLS certificate report). No shell access; outbound URL tools enforce SSRF protections.

Scrapes a URL and returns structured JSON: title, meta tags, headings, paragraphs, links, tables, extracted emails and phone numbers, and word counts. Accepts a single 'url' or a list of 'urls' for batch scraping. Includes retry logic and anti-detection headers.

Web content extraction and search. Get clean markdown from any URL or search results.

Extract readable text, metadata, headings, links and JSON-LD from one public static HTML page for research and retrieval.

Extract structured public business contact information from any public business website. Returns company name, emails, phone numbers, social profiles, postal addresses, and contact pages as structured JSON. $0.01 per request via x402 on Base.

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 l

Extract title, headings, paragraphs, links from any URL. Body: {"url": "https://example.com"}

Fetch a public https URL and extract HTML head metadata ($0.005 USDC via x402). https only — http, file, data, localhost, and private IPs are rejected. Caps: 1 MB body, ~10s timeout, max 5 redirects. Returns JSON { url, final_url, content_type, title?, description?, canonical?, lang?, open_graph, twitter, warnings }.

Extract entities/fields from text into a caller JSON Schema. Wikidata QID links + source spans. Fail closed (no charge) if output does not validate.

Extract text and tables from a Word .docx file ($0.005 USDC via x402). Input: a public https url to a .docx, or docx_base64. https only — http, file, data, localhost, and private IPs are rejected. Caps: 10 MB file, ~15s timeout, max 5 redirects. Tables appear as [table N] placeholders in text. Unicode dashes/minus signs normalized to '-'. Old binary .doc files are not supported. Returns JSON { fil

Extract links from URL or markdown; status + redirect + canonical; polite rate; honor robots.

Extract tables from a PDF ($0.01 USDC via x402). Pass pdf_base64 or a https pdf_url. Optional schema maps headers. Optional max_pages (1-10). Returns JSON { pages, tables: [{ page, headers, rows }], warnings }. Text PDFs are parsed in-worker; scans require HOSTED_PARSER_URL.

Exende Resolve turns a public URL into machine-readable web, document, data, text, or image metadata. It detects resource type, extracts static HTML, renders JavaScript pages only when needed, parses PDF text, JSON, CSV and plain text, and returns a normalized result.

Deterministic keyword extraction (term frequency, stopword filter). Pairs with url-to-markdown / condense. Amp port.

Prompt-injection scan for text your AI agent is about to read (web pages, emails, tool output, MCP tool descriptions): risk score 0-100, verdict, categories and exact spans. Detects instruction overrides, jailbreak personas, system-prompt extraction, fake chat delimiters, data-exfiltration URLs, tool hijacking, hidden Unicode and encoded payloads. Deterministic rules, no data stored.

Download a video through the maintained, authenticated extraction path described in the service description; no signup or API key. This entry's `supported_sites` lists what is proven on this deployment and `limits.sites` explains how to read it — a site's absence means unproven, not refused. Returns a zip by signed URL. This is the tool that is hard to replicate yourself.

Text-to-structured-JSON extraction: POST plain text, receive emails, phone numbers, dates, monetary amounts and URLs as JSON.

Fetch a public web page and return clean, readable Markdown (main-article extraction via Readability, scripts/nav/ads stripped). Body: {"url": "https://..."}. Private/localhost URLs are refused; pages over 2 MB are refused; output capped at 200k chars. Not charged on errors.

Search the web with AI-oriented result extraction.

Extract text from a digital PDF into page-numbered JSON with SHA-256 hashes. No account or API key. 0.005 USDC on Base per successful document via x402. Maximum 2 MiB and 25 pages. No OCR, summaries, URL fetching, or verification of document claims.