Yard Sale Gold — AI-powered yard sale scanner built with React and Cloudflare Workers
by wesbosTypeScript
Last 12 weeks · 4 commits
1 of 6 standards met
I added a sell mode to help my daughter and me sell our junk on Facebook Marketplace and at a yard sale. The scanner already did most of the research work; this points it at our own stuff and turns the result into prices and a listing we can use. What it does A Buy / Sell toggle on (remembered in localStorage). In sell mode the agent uses a seller prompt: Treats every clear subject of the photo (held up or set down) as the owner's item. No sale table or price tag needed, which the buy prompt requires. Keeps the same research flow (retail baseline, web search, eBay active listings), but prices against used-market evidence and returns: Marketplace price, with roughly 10–20% room to negotiate Lowest offer worth accepting Yard-sale price: a sticker price that sells the same morning, rounded, never above the lowest offer Listing title and description, honest about visible condition, with a Copy button Find criteria still apply in both modes. History gets All / Buy finds / My items filters, and accepts . Data changes Migration : Adds ( | , default , so existing rows become buy finds) plus nullable , , , , . Changes the unique index from to . and the worker's fallback matching are also scoped to the mode, so pricing your own copy of something you once spotted at a garage sale doesn't merge the two records. Deployments need before deploying. takes an optional form field (defaults to ) and returns a 400 for anything else before storing a frame or calling OpenAI. Also included (square 180px render of the existing icon) plus the `npm testtsc -b` is clean. Used locally and on a personal Cloudflare deployment from an iPhone. 🤖 Generated with Claude Code
Repository: wesbos/yard-sale. Description: Yard Sale Gold — AI-powered yard sale scanner built with React and Cloudflare Workers Stars: 83, Forks: 20. Primary language: TypeScript. Languages: TypeScript (79.6%), CSS (20%), HTML (0.4%). Open PRs: 1, open issues: 0. Last activity: 1w ago. Community health: 28%. Top contributors: wesbos.