MicroBuzz — Swarm Simulation Engine for Token Listing Intelligence

MiroFish (formerly MicroBuzz) is a swarm simulation engine that runs 50 AI agents across 5 behavioral clusters (degen, whale, institutional, community, technical_trader) to produce Expected Value predictions for token listing decisions. Part of Buzz BD Agent — the world's first autonomous exchange listing pipeline. Core innovation: 5 behavioral personas × 10 weight variations = 50 agents that independently evaluate a token. Their consensus feeds an EV formula (EV = P(success) × reward − P(failure) × cost) producing a mathematically-backed LIST, MONITOR, or REJECT decision. Post-hackathon evolution (Project Opus Brain): Migrated from external LLM calls to Claude Opus 4.6 as the brain running 24/7 via Claude Code on Hetzner. All 50 simulation agents now use rule-based verdicts — zero external LLM cost. The brain analyzes raw data endpoints directly. Connects to 25 intel sources including DexScreener, CoinGecko, OKX WebSocket, Helius, RugCheck, AIXBT, and Allium. Results displayed through cyberpunk-themed reports with animated MiroFish swarm visualization. Built through conversational AI collaboration — the human operator (Ogie) is a 20-year Executive Chef from Jakarta with zero CS degree. Every line of code written through Claude dialogue. Total infrastructure cost: $4.09/month. Live features: 135+ REST API endpoints, 55 database tables, 22 active data collection crons, Triple Verification data integrity, Sentinel v2.0 health monitor, ERC-8004 identity on 6 chains, Job Verification Receipts with SHA-256 hashes.