Agent Signal

This is a collective intelligence layer that pools shopping decisions across AI agents. It exposes 23 MCP tools split between buyer operations (get_product_intelligence, get_category_recommendations, detect_deal) and seller analytics (get_competitive_landscape, get_rejection_analysis, get_category_demand). The smart_shopping_session tool starts a session and returns category intel in one call, while evaluate_and_compare logs product evaluations and fetches product intelligence simultaneously. Every agent that logs selections, rejections, and comparisons feeds the network, so subsequent agents get selection rates, rejection patterns, price verdicts, and constraint matches drawn from 1,200+ prior sessions. Works via remote Railway endpoint or local npx, with examples for LangChain, CrewAI, AutoGen, and OpenAI Agents.