TariffMonkee

Pay-per-call web data APIs for AI agents over the x402 protocol. Real rendered-page fetching, structured extraction, search, and a multi-turn research orchestrator -- no accounts, no API keys.

Skills

• Content
  Use this when you need the readable text of a specific web page and either don't have your own scraper, or the page needs real JavaScript execution to render correctly (single-page apps, client-side-rendered docs, dynamically loaded articles) -- a plain HTTP fetch of the raw HTML would miss that content. Requires knowing the URL already; use tariffmonkee-search first if you don't. Returns Markdown, not structured data -- pair with tariffmonkee-extract if you need specific fields pulled out of the result rather than the whole page. Repeat calls for the same URL within 24 hours are served from cache, so re-checking a page you already fetched is cheap.
  contentbrowsermarkdownscraping
• Screenshot
  Use this when you need a visual capture of a page, not its text -- confirming what a page actually looks like, checking that a UI or layout change took visible effect, or producing visual evidence of a page's state at a point in time. The HTTP response body is the raw PNG image itself, not JSON. Renders with real JavaScript execution first, same as tariffmonkee-content, so client-side-rendered pages capture correctly rather than showing a blank or loading state.
  screenshotbrowservisualpng
• Pdf
  Use this when you need an archival or printable snapshot of a full page as a file -- a record of a page's content and layout at a point in time, suitable for saving, sharing, or printing. The HTTP response body is the raw PDF file itself, not JSON. If you only need the text content and not a file, tariffmonkee-content is cheaper and returns Markdown directly.
  pdfbrowserarchival
• Extract
  Use this when you already have raw text or Markdown content in hand -- from a prior tariffmonkee-content call, or text already in your own context -- and want specific structured fields pulled out of it against a JSON Schema you supply. Pure AI extraction with no browser render involved, so it's cheaper and faster than tariffmonkee-extract-url. Use tariffmonkee-extract-url instead if you only have a URL and haven't fetched the page content yet.
  extractionstructured-dataai
• Extract Url
  Use this when you have a URL but not the page content yet, and want structured data extracted from it in a single call -- fetches with a real rendered browser and extracts against your JSON Schema in one round trip, trying multiple AI models in sequence for reliability. If you already have the content in hand, tariffmonkee-extract is cheaper since it skips the browser render entirely. Results are cached for 24 hours per URL+schema combination, so repeating the same extraction on the same page is cheap.
  extractionstructured-dataaibrowser
• Aeo Audit
  Use this when you want to know how ready a page is for AI and agent consumption specifically, not human browsing -- it checks schema.org structured data, crawlability (robots.txt, meta robots, canonical tags), and llms.txt presence, then returns a 0-100 score with specific, actionable fixes. Useful for auditing your own site, a client's site, or deciding how much to trust a page as a citable source.
  auditseoaeoschema.org
• Search
  Use this when you need to find pages on a topic and don't already know a specific URL -- this is the entry point, not tariffmonkee-content, when you're starting from a question rather than a known page. Backed by three search providers with automatic failover, so a single provider outage doesn't take it down. Pass region=tw for a Traditional Chinese / Taiwan-region search lane instead of the global default. Returns titles, URLs, and snippets only -- pair with tariffmonkee-content for full page text, or use tariffmonkee-research instead if you want both search and fetch done for you in one call.
  searchdiscovery
• Research
  Use this instead of chaining tariffmonkee-search followed by several tariffmonkee-content calls yourself -- it does both in one call, deduping sources by domain and packing up to 10 rendered sources into a single roughly-8,000-token bundle sized for direct use in your context. Requires at least 3 successfully fetched sources to succeed, and reports sources_requested vs. sources_returned so you can see how many of the candidate set actually came back. Costs more than a single search or content call, but replaces several of them -- reach for this when you want a research bundle for a topic, not a single page or a single search results list.
  api
• Research Orchestrator
  Multi-turn research: checks your query is specific enough to search, automatically broadens and retries once if the first pass returns too few sources, and asks a clarifying question -- pausing the task, resumable with a follow-up message carrying the same taskId -- only if that isn't enough. Wraps the same research engine as tariffmonkee-research, with real iterative reasoning around it rather than a single fixed call.
  researchsearchorchestrationmulti-turnagentic

How to call

Listed in

Directories this entry was found in.