AntiBrow recipes

Task-level site adapters for AntiBrow. One file per command: an agent asks for reddit/hot and gets structured JSON back, instead of a browser handle and a scraping problem.

npm install -g anti-detect-browser

anti-detect-browser recipe update
anti-detect-browser recipe run reddit/hot --temporary --json
anti-detect-browser recipe run reddit/hot --profile shopper-01 --jq '.items[].title'

Platforms covered today

	commands
Search	google/search · duckduckgo/search
Marketplaces	amazon/search · walmart/search
Social & community	reddit/hot reddit/search reddit/comments · x/timeline · hackernews/top hackernews/item
Publishing	medium/tag
Developer	github/repo · pypi/package · npm/package
Listings & jobs	yelp/search · indeed/jobs
Identity checks	ipify/exit-ip · ipapi/exit-geo · antibrow/fingerprint

19 commands over 14 sites, including ones that answer a plain scraper with a captcha: Amazon, Walmart, Google, Medium, Yelp, Indeed. See what is published for exactly which ones have returned data here and which need a decent exit IP first.

Missing a platform? Ask for it. The sites with the strictest bot management are the ones this layer is for, and adding one is a pull request in this repo - not a release of the SDK. Open an issue with the site and the data you want, or write it yourself from GUIDE.md; an agent can usually do it in one sitting. Requests for platforms behind Cloudflare, DataDome, PerimeterX or Akamai are welcome: those are the interesting ones, and they are where a per-identity browser with its own exit IP is the difference between getting an answer and getting a challenge page.

One command, N identities

The part that is ours: the same command runs on N isolated identities at once.

anti-detect-browser recipe fanout ipify/exit-ip --profiles 'shopper-*' --concurrency 4

Four profiles, four personas, four cookie jars, four exit IPs, one command. If a site locks one of them, you discard a profile.

What a recipe is

export const meta = {
  id: 'reddit/hot',
  summary: 'Hot posts from the front page or one subreddit.',
  domains: ['www.reddit.com'],
  entry: 'https://www.reddit.com/',
  identity: 'any',
  args: [{ name: 'limit', type: 'number', default: 25, max: 100 }],
}

export async function run(ctx, args) {
  const res = await ctx.fetchJson(`/hot.json?limit=${args.limit}`)
  return { items: res.data.children.map((c) => ({ id: c.data.id, title: c.data.title })) }
}

No imports, no dependencies, no build step. run() executes inside the page, so relative fetches carry that profile's session for that site.

Safety

A recipe runs in a browser profile that may hold live logins, so the client treats it the way it would treat a browser extension:

• meta.domains is enforced at the network layer. A request to a host the recipe did not declare is blocked, so a Reddit recipe cannot reach your mail.
• Recipes are pinned by SHA-256 and only reviewed ones run by default. --allow-unreviewed opts in, and only on a --temporary profile.
• scripts/check.mjs is a convenience for reviewers, not the boundary. Read the recipe you are about to run; it is one screen of code.

What is published

19 commands over 14 sites. Everything marked reviewed has returned real data from a throwaway profile on this machine; recipe list marks the rest.

Reviewed	reddit/hot reddit/search reddit/comments hackernews/top hackernews/item github/repo pypi/package npm/package duckduckgo/search medium/tag amazon/search walmart/search ipify/exit-ip ipapi/exit-geo antibrow/fingerprint
Not reviewed yet	google/search indeed/jobs yelp/search x/timeline - see below

The unreviewed ones need --allow-unreviewed and a --temporary profile. That is the point of the flag: a recipe nobody has seen work should not run against a profile you keep.

The unreviewed four need a decent exit IP

They are written and their challenge detection is verified, but none of them returned data here from a cold throwaway profile on a flagged exit:

recipe	what stopped it	what it needs
google/search	consent wall, intermittently - the same recipe got a clean result page on one run and a wall on the next	a clean residential exit, and a profile that has accepted consent once (so run it on a profile you keep, not --temporary)
indeed/jobs	Cloudflare interstitial	a clean residential exit
yelp/search	anti-bot challenge (the vendor's challenge frame is on a host the allowlist blocks, so it can never be solved in-page)	a clean residential exit
x/timeline	not signed in	a profile signed in to x, ideally on the exit that account is normally used from

This is not something a recipe can fix. Exit reputation, request cadence and the account history a profile carries decide whether these sites answer at all, and none of the three is a property of the parsing code. Every one of them reports the challenge explicitly rather than returning an empty list, so on a fanout you can see which identity got stopped.

Contributing

Read GUIDE.md - or have your agent read it:

anti-detect-browser recipe guide

Then:

anti-detect-browser recipe scaffold <site>/<command>
node scripts/check.mjs --fix
anti-detect-browser recipe test <site>/<command>

One site per pull request. Explore the site from a throwaway profile, never your own account.

License

MIT. See LICENSE.