Technical Documentation

This sub-section is the implementation-level reference for engineers. Where Architecture explains what the system is, these pages explain how it behaves in code.

What’s here

• Schema-Driven Model — how networks, domains and items are defined by schemas, and why nothing domain-specific is hardcoded.
• Read & Write Paths — the two fetch layers (instance-local vs. inter-instance), discovery, caching, and how signals are written.
• Tech Stack — the full toolchain, runtimes and conventions across both DPGs.

Cross-cutting principles

A few principles hold across the entire codebase:

• Configuration as code. No domain- or environment-specific value is hardcoded. Values are read once at startup from a config loader or env, with per-environment overrides.
• Typed boundaries. Service methods return a Result<T, BaseError> rather than throwing across boundaries; routes return machine-readable error codes.
• Schema-first. Forms, cards and API contracts derive from Zod/JSON schemas, so domains evolve without code changes.
• Observability by default. Log entries carry operation, status, latency_ms for external calls, and error/error_type on failure. No bare console.log in module code.
• Tested to a bar. Vitest only; cross-package tests use the in-memory fake from each package’s ./testing export; target ≥ 70% line coverage.