Back to tools
5 endpoints

Search1API

Live web access for agents through one schema — search that fuses Google, Bing, DuckDuckGo, Yahoo and Yandex or targets one engine, the Chinese web (Baidu, Bing CN, 360, Quark, WeChat articles, Bilibili), and Reddit, X, GitHub, YouTube, arXiv and Wikipedia; news from Google, Bing, Reuters and Hacker News; the full text of the top results in the same call; single-page reads to clean Markdown with dedicated parsers for major sites; sitemap discovery; and GitHub and Hacker News trending.

Part of Search & SEO

Get started by giving this to your agent
$Set up https://monid.ai/SKILL.md, and then use Monid to show me what I can do with Search1API.
Search1API
Read Page as Markdown
/crawl

Fetch one public URL as clean readable text.

$0.0010per call
0 Markup
Search1API
Search the Web
/search

Search the live web across 18 engines, optionally with each top page's full text.

from $0.0010per call
0 Markup
Search1API
Get Trending Topics
/trending

List what is trending right now on GitHub or Hacker News.

$0.0010per call
0 Markup
Search1API
List Site URLs
/sitemap

List the URLs a site's sitemap declares, or every link on one page.

$0.0010per call
0 Markup
Search1API
Search News
/news

Search recent news articles, optionally with each top article's full text.

from $0.0010per call
0 Markup

Try it with your agents

01
Map a product's competitors
Set up https://monid.ai/SKILL.md, and then use Monid to search the web for the strongest alternatives to Notion and build a comparison table from the top sources.

How-to guides

Which local businesses still have no website?businesses with no website, screened for the chainssearch google maps for <trade> in <city> with litescrape on monid and tell me how many of the results already have a website and a rating above 4.5. if almost all of them do, say so and stop before i build a list the answer: forty results and not one of them had a missing website the ratings: nineteen of the first twenty sat at 4.5 or above it degrades as you page: the first page carried a website and a rating on all twenty rows, but across 75 unique rows the website was present on 69 and the rating on 72, so the deeper you go the more rows the screen cannot judge the verdict: the category is saturated, so a no-website lead list here would be empty what it cost: three hundredths of a cent to avoid building the wrong listWhere does my page actually rank?the live page, not what the tool says about ituse monid to pull the live google results for <keyword> with litescrape, and give me the organic results with their urls. if you want what the ai overview cites, call the dedicated ai-overview endpoint as well: the search response carries the overview text and not one of its links the page: nine organic results, HubSpot first and Mailgun second the absence: neither the subject domain nor its competitor is anywhere in the top nine the overview has no links: the search response does carry an ai overview block, and on a second keyword it held text, snippets and highlights with zero urls in it; the citations come from a separate endpoint at the same price, which returned nine references ten is not on offer: asking for ten results returned eight, and asking for a hundred returned the same eight, so write the sentence around what is on the page rather than a count the comparison: the cheapest paid endpoint that answers this is $0.06 a result, 400x the priceDo AI answers cite my brand?and why there is no single share-of-answers numberask google ai mode <question> with litescrape on monid and give me the answer plus every source it cites, in order. check its health flag first and budget three attempts on a 503, not one: on our second run it took two failures before a 200 what came back: the generated answer plus ten cited references, in the order the answer used them the price: $0.00015 against $0.01 for the same surface, which is 66x the catch: it was flagged degraded on both runs and 503'd on both; once a single retry cleared it and once it took three attempts, so one retry is not a budget a 503 still reads as COMPLETED: the run record shows a completed status with the 503 inside it, so a script filtering on status ingests the failure silently; read the provider's own status code the answer varies: answers are generated per request, so two calls the same day differ