Skip to main content

​
Architecture Overview

This section is the visual reference for how CONA is built. Diagrams are organised as C4 levels: each level zooms in one step, so you can stop at the depth you need.
Every diagram in this section is derived from the code, not from intent. Edges are verified against actual imports and call sites. If a diagram disagrees with the code, the diagram is the bug — see Conventions.
The set is complete: 20 diagram pages, 43 diagrams, every one validated against Mermaid’s parser, measured for legibility at the docs column width, and traced back to source files.

​
How to read this section

​
L1 — Context

System Context

The single high-level picture: users, applications, infrastructure, and external providers.

Data Flows & Trust Boundaries

Where customer and financial data travels, which boundaries it crosses, and which subprocessors touch it.

​
L2 — Containers

Deployment Topology

Vercel, Fly.io, Temporal Cloud, Supabase, and Redis — what runs where.

Package Graph

The monorepo’s applications and packages, their real dependency edges, and the layering rules.

​
L3 — Applications

Webapp

Route groups, server actions, library layers, and API ingress.

Portal

The customer-facing document portal.

Console

The superadmin console.

Temporal Workers

Worker bootstrap, task queues, and activity registration.

​
L3 — Business domains

Core Domain Map

The domains inside @cona/core, grouped into bounded contexts.

Accounting Engine

Posting matrix, rules engine, general ledger, job queue, and periods.

Document Lifecycle

From creation through formats and delivery to the customer portal.

Reconciliation & Payments

Payment matching, bank handoffs, and the reconciliation agent.

Data Model

The Prisma schema as domain clusters rather than one diagram.

​
L3 — Cross-cutting concerns

Integrations & Adapters

The adapter registry and every external provider CONA syncs from.

Ingress: OAuth & Webhooks

How external systems get in — callbacks, webhooks, and the public API.

Temporal Orchestration

Workflow to activity to core call chains, retries, and schedules.

Auth, Tenancy & Actors

Auth0, organisation isolation, and mutation attribution.

Observability

Axiom pipelines, OpenTelemetry, Sentry, and sampling.

Redis Caching

Cache key model, version prefixing, counters, and locks.

Copilot Agents

Mastra, the CopilotKit runtime, and agent tooling.

​
Related reading

  • CODEBASE_MAP.md at the repository root — text-first navigation of the same territory
  • Temporal Workflow Architecture — deeper narrative on the ingestion and accounting phases
  • Core Principles — the rules these diagrams should reflect