TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Show HN: AI web scraper (no selectors, $2/1k pages, built by 1 dev)

4 点作者 zeeb0t6 天前
Hi HN - I’m the solo founder of InstantAPI, a web scraping API that turns any webpage into structured JSON without HTML selectors or coding. I built it because I was frustrated with scraping tools that required setup, CSS parsing, or came with confusing pricing and limits.<p>InstantAPI is designed to be radically simpler and more scalable:<p>- You send a URL and the data structure you want (via example or JSON schema) - it returns just that, parsed by AI.<p>- It handles JavaScript, CAPTCHAs, and rotating proxies automatically.<p>- You can &#x2F;scrape any page, extract &#x2F;links by description, automatically find &#x2F;next page in paginated results, and &#x2F;search Google SERPs.<p>- No per-feature pricing - just a flat $2 per 1,000 fully scraped pages, including all rendering and AI parsing.<p>- It includes a no-code protocol (MCP) that lets LLM agents request structured data directly, without wasting tokens on markdown or full page content.<p>I built this alone using AI tools and bootstrapped it all - and in testing it’s outperformed a few VC-backed tools that charge 10x more or require selector logic and scraping scripts.<p>Would love your thoughts, questions, or feedback. Happy to share tech details or walk through how it works. Thanks!

1 comment

cepc9125 天前
Sounds interesting. Which AI are you using to generate the code? Or is it one you built yourself. Can you confirm the accuracy of all the generated code?