Screen second-hand marketplace listings against what you want
A browser agent reads second-hand listings and Jev decides which match your requirements, so you only open the good ones.
Author's figure: Author's figure: roughly 26 listings read per minute, 406 ms and $0.00085 per decision (Alan Daitch). Reported by the builder, not measured by us.
The use case
Buyers hunting for used gear, and resellers sourcing stock, scroll through hundreds of listings with vague titles. Here Playwright opens each listing, the text and price go to Jev with typed questions built from your requirements, such as right model, acceptable condition, price under a cap, seller located nearby, and only matches are saved. The person reviews the shortlist, contacts the seller and inspects the item before paying. Jev should not be trusted with the price arithmetic itself; compare prices in code and keep Jev for the reading.
Example workflow
- Write requirements as yes/no questions
- Use Playwright to open search results on the marketplace
- Send each listing's text to Jev and check price limits in code
- Save matches with links to a Google Sheet or Telegram alert
- The person contacts sellers and inspects items before buying
Tools
- Jev (TypeSafe AI)
- Playwright
- Google Sheets
- Telegram
Risks and guardrails
- Marketplaces may forbid scraping
- Listing photos can contradict the text
- Scam listings can pass text checks
Sources
Written by Agents Lookup. Credits and official pages for the tools this use case names:
- A second-hand shopping agent by Alan Daitch on Made with Jev: madewithjev.com/builds/second-hand-shopping-agent
- A second-hand shopping agent: original post or repo by Alan Daitch: x.com/AlanDaitch/status/2100757989212754085
- Playwright MCP server (official page): github.com/microsoft/playwright-mcp (checked 2026-11-09)