Haunt

Web extraction for AI agents. Send a URL and a plain-English prompt, get clean structured JSON or Markdown from permitted public pages. When a page cannot be read it returns a typed reason instead of invented data, and the failed call is not charged.

Skills

• Extract structured data from a web page
  POST /v1/extract with a url and a prompt. Returns JSON matching your prompt, or a typed failure. Blocked pages cost nothing.
  web scrapingextractioncrawlstructured data
• Read a page as Markdown, no key needed
  GET /r/<url> returns clean Markdown. Rate limited, never paywalled.
  readermarkdown
• Read Reddit without Reddit API credentials
  POST /v1/reddit for posts, /v1/reddit/comments for a thread. Takes a subreddit URL, a post URL, or shorthand like r/webdev. No app registration and no OAuth. Costs 1 credit, failures free.
  redditsocialforumsno api key
• Build a company profile from its website
  POST /v1/company/enrich with a url. Returns what the company does plus the contact and social details published on the site, read live rather than from a stale database. Costs no credits.
  enrichmentcompany datasalesresearch
• Get public GitHub repository details
  GET /v1/github/repo?owner=<owner>&repo=<repo>. Description, stars, forks, language, licence, topics, last push. Costs no credits.
  githubrepositoryopen source
• Extract many URLs in one call
  POST /v1/extract/batch with a list of items and one prompt. Reports how many succeeded and the credits used.
  batchbulkcrawl

How to call

https://hauntapi.com

Listed in

Directories this entry was found in.