I built this using Claude Code as an orchestration layer — parallel research agents deployed per resource and per country, each producing a self-contained deep

I built this using Claude Code as an orchestration layer — parallel research agents deployed per resource and per country, each producing a self-contained deep dive with sourced data. Cascade models then identify how disruptions compound across domains simultaneously. The motivation: most war analysis is top-down political commentary. But wars are resource disruption events. If you model from physical constraints upward — who produces what, what ships through where, what breaks when supply X goes offline — you get different predictions than pundit-level analysis. Findings that surprised me: 1. The chip famine is three independent input crises hitting simultaneously (helium 33% offline + bromine 67% at risk + neon combinatorial) — no single-source fix exists 2. The food crisis was locked in by March. Northern Hemisphere spring planting happened with disrupted fertilizer supply. Even an immediate ceasefire can't fix Q3-Q4 yields. 3. November 2026 is a convergence point where US midterms, China's gallium/germanium export deadline, and Europe's winter energy crisis all collide. 4. Insurance is a bigger weapon than missiles. The $1:$390 trade multiplier means Lloyd's war-risk designations freeze far more economic activity than physical damage. You can chat with the full research via 1. NotebookLM (link above), 2. browse the site (https://hrishirc.github.io/iran-war-2026-analysis/ https://hrishirc.github.io/iran-war-2026-analysis/), 3. or dig into the source (https://github.com/hrishirc/iran-war-2026-analysis https://github.com/hrishirc/iran-war-2026-analysis). All sourced from major news outlets, think tanks, and government reports (Feb-March 2026). Built with Claude Code using an "Agentic Brain" knowledge architecture — the CLAUDE.md in the repo documents how 76 files are organized and interconnected.