I made the open protocol layer behind my B2B supplier matching project public
I’ve been building MapleBridge, a sourcing workflow for North American buyers looking for better-fit Chinese manufacturers.
One thing I kept running into: supplier search is not only a search problem. A buyer can have a very specific brief — product, MOQ, target market, packaging, certification path, timeline — but most supplier directories still reduce that into keywords and listings.
So I separated part of the work into an open layer.
MapleBridge Open documents how I think about:
- buyer intent
- supplier capability
- match explanation
- What still needs human review before an introduction
It does not include customer data, private ranking logic, crawler sources, or production prompts. It is more like a public protocol layer for AI-to-AI supplier search.
Open docs:
GitHub:
https://github.com/jinjihuang88-ui/maplebridge-open
I’d be interested in feedback from anyone thinking about agent workflows, marketplaces, or structured matching.

Replies