98 lines (80 loc) · 2.05 KB

ARC-Core Architecture Overview

Core principle

ARC-Core is designed around a simple doctrine:

models may assist, but canonical state and execution authority stay inside structured infrastructure.

Major layers

1. API layer

arc/api/main.py
arc/api/routes.py
arc/api/deps.py

Responsibilities:

application startup
route exposure
UI mounting
manifest/health surface
auth/session endpoints
analyst and operator workflow endpoints

2. Core layer

arc/core/config.py
arc/core/db.py
arc/core/auth.py
arc/core/schemas.py
arc/core/risk.py
arc/core/simulator.py
arc/core/util.py

Responsibilities:

config and paths
SQLite schema and connection bootstrap
role gates
typed input/output models
risk banding
lightweight simulation primitives
normalization and event fingerprinting

3. Service layer

arc/services/ingest.py
arc/services/graph.py
arc/services/cases.py
arc/services/watchlists.py
arc/services/proposals.py
arc/services/notebook.py
arc/services/audit.py
arc/services/authn.py
arc/services/connectors.py
arc/services/bootstrap.py
arc/services/geospatial.py
arc/services/resolver.py

Responsibilities:

business logic
ingest and graph updates
analyst workflow
audit/receipt logic
session logic
connectors
geospatial processing
demo bootstrap and seed data

4. Geo layer

arc/geo/geometry.py
arc/geo/estimator.py

Responsibilities:

polygon / structure geometry helpers
estimated location and spatial confidence helpers

5. UI layer

arc/ui/*.html

Responsibilities:

dashboard-level views
graph/timeline/case/geospatial pages

6. Tests

tests/*.py

Responsibilities:

smoke coverage
workflow coverage
geospatial coverage
hardening checks
v6 operations checks

Architectural value

This architecture already shows the intended pattern:

routes stay thin
services hold workflow logic
core defines the invariants
receipts and proposals are explicit
future model layers can plug in without owning truth