Tools Decisions

Purpose

This page documents the deliberate tool choices for FactHarbor POC → Beta.
The goal is low friction now, no rewrites later, and clear upgrade paths.

Phase 1 — POC1 (Local, Low Infrastructure)

Core Principles

• Minimize setup and cognitive load
• Avoid infrastructure that is not strictly required
• Keep architecture identical to Beta where possible

Selected Tools

Backend

• ASP.NET Core (.NET 8)
  • Role: system of record (jobs, status, results, events)
  • Reason: strong typing, long-term maintainability, C# expertise

• SQLite (EF Core provider)
  • Role: local persistence
  • Reason: zero setup, file-based, easy Postgres transition

Frontend & Orchestration

• Next.js (TypeScript)
  • Role: UI + AI orchestration
  • Reason: fast iteration, strong ecosystem, good AI SDK support

• Vercel AI SDK
  • Role: LLM abstraction layer
  • Reason: OpenAI-first, Claude-ready, clean provider switching

AI Providers

• OpenAI (default)
• Anthropic (optional / later)
  • Selection via environment variables

Editors

• Visual Studio 2022 — primary for .NET API
• Cursor or VS Code — primary for Next.js / TypeScript

Phase 2 — POC → Beta Transition

Changes (No Rewrite)

• SQLite → PostgreSQL (EF Core provider switch)
• Local run → hosted environments
• Add authentication, quotas, persistence hardening

Additional Tools Introduced

• Docker (optional, for DB parity)
• Managed Postgres
• CI (GitHub Actions)
• Hosting
  • Next.js → Vercel
  • .NET API → Azure App Service / Container Apps

Non-Goals (Explicitly Deferred)

• Kubernetes
• Microservices split
• Complex queue systems
• Multi-region deployment

These are intentionally deferred to reduce early risk.

Key Decision Summary

• Architecture is stable from POC1 onward
• Infrastructure is phased, not front-loaded
• Tooling choices prioritize clarity over novelty