Ready-made agents for the jobs people do by hand
CrawlGenie Agents are web agents that need no coding: pick one, fill two fields, get a table. Greenhouse jobs, Shopify catalogues, SEC filings, arXiv papers, Google News, App Store results, any pricing page, any list on any page. None of the 61 needs a model key. When a job is unique, describe the goal and CrawlGenie drives a browser with your model and returns exactly the fields you asked for.
Zero model tokens for the 61
They read structure, not prose: rows and CSV, repeatable, honest when a site blocks them.
Goal in, JSON out
Give a goal and the shape of the result; run it on 200 sites in a batch with a concurrency cap.
Built for the dashboard
Every agent has a form, a result table, a CSV button, and a "turn into a monitor" button.
What happens when you use Agents?
Choose
Browse by industry: local and leads, e-commerce, jobs, reviews, social, news, SEO, finance, research, government, legal, travel, events.
Run
Fill the inputs (a URL, a query, a ticker). Results arrive as a table in seconds.
Keep going
Download CSV, schedule it as a monitor, send rows to Memory, or call it from code as one tool.
Verify it’s you
Enter the 6-digit code from your authenticator app.
{{totp:acme}} · the seed never leaves the server| Invoice | Date | Amount | Status | |
|---|---|---|---|---|
| INV-2291 | Oct 3 | $4,120.00 | Paid | Download PDF |
| INV-2297 | Oct 9 | $980.00 | Due | Download PDF |
| INV-2304 | Oct 14 | $2,375.00 | Due | Download PDF |
What do people use Agents for?
Recruiting
New roles across Greenhouse, Lever, Ashby and job boards every morning.
Sales and growth
Enrich a domain list with contacts, tech stack and brand assets; track competitor pricing pages.
Finance and legal
Filings, Federal Register notices, court opinions, quotes and FX rates as rows.
What you get, in detail
The 61 built-in agents cover the jobs people otherwise do by hand or pay a scraper vendor for: job boards (Greenhouse, Lever, Ashby, RemoteOK), stores (Shopify, WooCommerce, eBay), reviews, places and contacts, social posts and mentions, news, SEO signals, stock and crypto quotes, filings and public records, papers and code repositories, events and travel listings, any pricing page, and any list on any page. They read structure rather than prose, so none of them needs a model; they return rows with stable columns, report how many pages they read, and say plainly when a site blocked them or needs a proxy.
When a job is unique, a goal agent takes over: describe the task in plain English, give a starting URL and the shape of the result you want, and CrawlGenie drives a browser with your own model key, reading ranked snapshots instead of page dumps, and returns JSON in that shape. Batches run the same goal across up to two hundred URLs with a concurrency cap; runs can be synchronous, asynchronous or streamed, and every run carries its input, output and observation token counts. Any agent can be turned into a monitor with one click.
For developers
POST /v1/extractors/{id}/run for rows; POST /v1/agent/run {task, url, schema} and /v1/agent/batch for goal agents with sync, async or streamed results. Web agents · Library.
What would this save you?
Move the sliders. The saving is computed from the measured ratios above and the model price you pick.
Assumes 10% of the page is kept per read (measured range 5–15%). Receipts in the product show the real number for your traffic.
Questions people ask
Do I need an OpenAI or Anthropic key?
Not for the 61 built-in agents. Goal agents use your own model key, stored once; there is no markup on tokens.
What if a site blocks the agent?
The result says so instead of returning an empty table. A few sites need a residential proxy; the library labels them.
Spend tokens on answers, not on pages.
Create a workspace in a minute. Connect ChatGPT, Claude, Cursor or your own agents. Every answer comes with a token receipt.