Hardening program: architecture, data integrity, security, performance, CI, and observability

[Occumed79]

Goal

Track the remaining Network Map hardening as a coordinated program rather than one oversized PR.

Existing workstream

• Complete the full application UI hardening and ownership consolidation #168 — UI and runtime ownership consolidation (highest priority)

Additional workstreams

• Provider search architecture consolidation
• Coordinate and provider-data integrity
• Upload pipeline and rollback hardening
• API authorization and abuse protection
• Database consistency and lifecycle hardening
• Large-map rendering and memory pressure
• Network resilience and external-service isolation
• Real test coverage and deployment gates
• Observability and diagnostics

Execution order

Wave A — eliminate competing truth/ownership

1. Complete the full application UI hardening and ownership consolidation #168 UI/runtime ownership
2. Provider search architecture
3. Coordinate/provider-data integrity

Wave B — make writes and persistence safe

4. Upload pipeline/rollback
5. API authorization/abuse protection
6. Database lifecycle/consistency

Wave C — make runtime behavior resilient at scale

7. Large-map rendering/memory
8. External-source/network resilience
9. Observability/diagnostics

Wave D — enforce everything continuously

9. Real browser/API test coverage and deployment gates

The CI workstream may add scaffolding early, but production gates should only become required once the preceding architecture changes are stable.

Program rules

• One focused PR per workstream unless a dependency requires otherwise.
• Do not combine broad UI, search, schema, security, and rendering rewrites in one branch.
• Preserve production behavior while migrating consumers.
• No destructive schema/table deletion until reconciliation and migration coverage exist.
• No direct production-data mutations as part of tests.
• Every PR must include rollback notes and exact validation results.
• Production acceptance must verify the exact deployed revision, not merely a responding URL.

[Occumed79]

Workstreams are now split into focused issues so they can be implemented and validated without one unsafe mega-PR:

• Complete the full application UI hardening and ownership consolidation #168 — UI/runtime ownership consolidation
• Hardening 2: consolidate provider search into one authoritative backend pipeline #171 — Provider search architecture consolidation
• Hardening 3: enforce coordinate quality and provider-data integrity #172 — Coordinate/provider-data integrity
• Hardening 4: add upload preview, staging, idempotency, and rollback #173 — Upload preview/staging/idempotency/rollback
• Hardening 5: strengthen API authorization, CORS, rate limits, and write auditing #174 — API authorization/CORS/rate limits/write auditing
• Hardening 6: enforce database lifecycle, schema consistency, and integrity constraints #175 — Database lifecycle/schema consistency/integrity
• Hardening 7: keep large provider maps responsive without clustering #176 — Large-map rendering/memory without clustering
• Hardening 8: isolate external provider sources with timeouts, cancellation, and circuit breakers #177 — External-source/network resilience
• Hardening 9: add real browser/API test coverage and deployment gates #178 — Real browser/API test coverage and deployment gates
• Hardening 10: add end-to-end observability and diagnostics export #179 — End-to-end observability and diagnostics export

Dependency order remains: #168 → #171 → #172; then #173/#174/#175; then #176/#177/#179; #178 is developed throughout and becomes the final enforcement gate.