What is PKP?

PKP (Product Knowledge Protocol) is an open standard for representing product knowledge in a way that AI agents can easily consume, compare, and reason about.

The Problem

Today's e-commerce data is:

• Scattered across different platforms and formats
• Inconsistent in structure and terminology
• Hard to compare across categories and brands
• Difficult for AI to reliably parse and understand

The Solution

PKP provides:

1. Standardized Format - PRODUCT.md files with YAML frontmatter
2. Category Schemas - Consistent specs per product category
3. Confidence Metadata - Source attribution and verification
4. Web-native Distribution - Data lives at /.well-known/pkp/

Architecture Overview

Layer 0 - Static Files (on vendor domains)
├── /.well-known/pkp/catalog.json   → Product index
├── /.well-known/pkp/products/*.md  → PRODUCT.md files
└── Access: HTTP GET, zero cost

Layer 1 - Catalog MCP Server
├── Serves local PKP catalog via MCP
├── Tools: search, compare, filter
└── For: Single catalog access

Layer 2 - Registry MCP Server
├── Indexes multiple PKP catalogs
├── Cross-domain product search
└── For: Global product discovery

Key Concepts

PRODUCT.md

Each product is described in a Markdown file with YAML frontmatter:

---
schema: pkp/1.0
sku: "macbook-air-m4"
brand: "Apple"
name: "MacBook Air M4"
category: "notebooks"
summary: "Ultra-thin laptop with M4 chip"

specs:
  screen_size: 13.6
  processor: "Apple M4"
  ram_gb: 16
  storage_gb: 512
---

## Overview
The MacBook Air M4 is Apple's thinnest laptop...

Category Schemas

Each category has defined specs that enable comparison:

Category	Key Specs
smartphones	display_size, processor, ram_gb, camera_mp
notebooks	screen_size, processor, ram_gb, gpu
tvs	screen_size, resolution, panel_type

Confidence Levels

Data sources are ranked by reliability:

1. manufacturer - Official vendor data
2. retailer-feed - Authorized retailer feeds
3. community - User-contributed data
4. ai-generated - AI-extracted data
5. scraped - Web-scraped data

Use Cases

• AI Shopping Assistants - Compare products intelligently
• Product Catalogs - Structured data for e-commerce
• Price Comparison - Aggregate prices across retailers
• Review Aggregation - Consolidate product reviews