LocalMarket AI — C2C Marketplace & BU Bargain Finder

Inspiration & Problem Statement

Buying pre-owned goods on local peer-to-peer (C2C) platforms like Facebook Marketplace or OLX is notoriously tedious. Buyers spend hours manually filtering locations, calculating physical distances for COD (Cash On Delivery) meetups, and checking whether sellers are willing to negotiate or urgently selling ("BU / Butuh Uang").

LocalMarket AI bridges this gap using WebMCP (Web Model Context Protocol). By exposing structured tools directly to AI agents inside browser contexts (such as ChatGPT in-app browser or Chrome with WebMCP enabled), an AI agent can instantly assist users in finding nearby pre-owned items, filtering urgent deals, and negotiating prices.

What It Does

Distance Radius Filtering: Filters pre-owned items by physical COD distance (e.g. < 1 km, < 3 km, < 5 km).
⚡ BU (Butuh Uang) Deals Discovery: Instantly isolates urgent secondhand sales sold below standard market value.
WebMCP Price Bargaining: Exposes native make_offer tools allowing AI agents to negotiate prices and propose COD meetups automatically.
Interactive WebMCP Inspector: Built-in visual overlay displaying real-time tool registrations and executions.

WebMCP Implementation

LocalMarket AI implements window.document.modelContext.registerTool({...}) for 5 core tools:

search_nearby_deals: Filter listings by query, location radius (km), budget, and category.
filter_urgent_bu: Extract top urgent "BU (Butuh Uang)" bargains.
get_item_details: Retrieve seller verification score, item condition rating, and meetup locations.
make_offer: Submit custom price bargain offers and COD requests to sellers.
view_my_offers: Retrieve active user/agent offer shortlists.

How We Built It

Frontend: Clean, responsive dark-themed UI built with HTML5, Tailwind CSS, and Vanilla ES6 JavaScript.
Protocol Integration: Pure client-side WebMCP implementation with graceful degradation for standard browsers.
Dataset: Pre-owned item catalog enriched with geolocation metadata, seller verification ratings, and COD safety scores.
Testing: Fully tested with a 17-point automated blackbox and functional test suite (test_runner.js).

What's Next for LocalMarket AI

We plan to expand LocalMarket AI by integrating real-time geolocation APIs, web push notifications for accepted bargain offers, and multi-agent automated price negotiations.