Author: Harshit Anand
Built on Neuro-SAN as the sole multi-agent orchestrator. Source problem statement: ../hackathon-delivery-intelligence.md.
| # | Document | Purpose |
|---|---|---|
| 01 | Proposed Solution | Master specification — the single source of truth: problems, solution, agent network, risk model, trust ladder, contracts, scope. All other documents derive from it. |
| 02 | Data Flow Diagrams | DFD Level 0 (context), Level 1 (system), Level 2 (drill-downs) + data dictionary + flow invariants. |
| 03 | High-Level Design | Quality attributes, logical/component/runtime views, deployment (compose + Kubernetes), security, scalability, availability, observability, ADRs, risks. |
| 04 | Low-Level Design | Project layout, full agent-network HOCON, JSON contracts, all 19 coded tools, LLM config (NVIDIA NIM), Gateway API + CI/CD adapters, PostgreSQL DDL, dashboard, deployment artifacts, testing, error handling. |
| 05 | Architecture Diagrams | Six views: landscape, containers, agent network topology, cross-stage signal flow, production K8s, hackathon compose. |
| 06 | Frontend Design | Decision Dashboard SPA: stack (React 19 + Vite 7 + TS + Tailwind v4 + React Router; plain fetch + native EventSource), routes, component hierarchy, data/SSE layer, design system, auth gating — derives from 01 §14 and 04 §9. |
| 07 | Implementation Plan | Contract-first, component-based build plan: tracer bullet, tracks A–D, milestones M0–M5, 3-day solo schedule, risk register. |
Reading order: 01 → 05 (visual overview) → 03 → 02 → 04 → 06. Diagrams are Mermaid — render natively on GitHub and in VS Code (Markdown Preview Mermaid extension).