pn-new-crm/02-FEATURE-INVENTORY.md

Port Nimara CRM — Feature Inventory: Keep / Cut / Rethink

Compiled: 2026-03-11

Each feature is categorized as KEEP (carry forward as-is or with minor cleanup), RETHINK (keep the functionality but redesign the implementation), or CUT (drop entirely from the rebuild). A rationale is provided for every decision.

---

1. Core Domain Features

Feature	Verdict	Rationale
Interest CRUD (create, read, update, delete)	KEEP	Heart of the CRM. Whitelist filtering and audit logging are good patterns.
Interest list with search, filter, sort	RETHINK	Keep functionality. Replace unbounded fetch (limit: 1000) with proper cursor-based pagination and server-side search.
Interest detail view with tabs	KEEP	Core UI pattern. Rebuild with Maritime-only components.
Sales Process Level pipeline (8 stages)	KEEP	Proven workflow. Model transitions explicitly (state machine) instead of raw field mutation.
Frontend auto-promotion (yacht dims → Specific Qualified Interest)	RETHINK	Business rule should live server-side only. Currently duplicated in frontend and backend — single source of truth needed.
Berth CRUD and specifications	KEEP	Essential. Imperial + metric pairs are correct for the marina domain.
Berth area/status filtering	KEEP	Core navigation pattern for sales team.
Berth-to-interest linking/unlinking	KEEP	Critical workflow with auto-status rules. Keep the business rules, improve with database-level constraints.
Berth auto-status (Available → Under Offer on link)	KEEP	Essential business rule. Enforce at database/service layer, not scattered across endpoints.
Berth status override mode	KEEP	Manual override capability is important for edge cases.
Berth recommendations (separate from committed links)	KEEP	Useful sales workflow distinction. Implement as proper join table.
Duplicate detection (interests)	RETHINK	Useful feature. Replace Levenshtein blocking approach with database-level fuzzy matching (pg_trgm). Simplify scoring.
Duplicate merging (interests)	RETHINK	Keep merge capability. Add preview/dry-run mode. Improve rollback handling (currently fragile).
Duplicate detection (expenses)	KEEP	Same approach as interests. Rebuild with improved matching.

2. EOI / Signature Features

Feature	Verdict	Rationale
EOI document generation via Documenso	KEEP	Critical business process. Consolidate two duplicate ~400-line implementations into single EOIService.
3-party sequential signing (Client → Developer → Sales)	KEEP	Core legal workflow. Preserve exact signing order and notification chain.
EOI signature tracking	RETHINK	Keep tracking. Replace 15+ nullable text fields on Interest with dedicated signature_events table.
Documenso webhook handler	KEEP	Deduplication via signature hashing and locking is well-designed. Extract into testable service module.
Background notification processing (30s poll)	RETHINK	Keep the queued notification pattern. Replace Nitro experimental task with BullMQ job. Reduce to event-driven rather than polling.
Fallback signature polling (5-min)	KEEP	Important resilience pattern for missed webhooks. Keep as scheduled job.
EOI reminder system (morning/afternoon)	RETHINK	Currently disabled in production. Re-evaluate need with users. If keeping, implement with proper job queue.
Manual EOI upload (bypass Documenso)	KEEP	Business need for pre-signed documents. Keep the immediate status promotion.
EOI document validation (Documenso sync check)	RETHINK	Manual-trigger only. Automate as periodic reconciliation job.
EOI delete/cleanup	KEEP	Needed for error recovery. Ensure it cleans up all three stores (NocoDB, Documenso, MinIO).
Signing link QR codes	KEEP	Useful for in-person signing scenarios at the marina.
Embedded signing links (iframe-based)	RETHINK	Evaluate if direct Documenso links are sufficient vs. embedding in CRM.

3. Email Features

Feature	Verdict	Rationale
Email composer with HTML signature	RETHINK	Keep functionality. Replace raw Nodemailer with transactional email service (Resend/SendGrid) for outbound. Extract HTML templates to template files.
Email thread viewer (IMAP sync)	RETHINK	Keep functionality. Single IMAP implementation (eliminate duplicate pool/standalone). Move sync to background worker. Store thread metadata in database, not just MinIO JSON.
Email attachment support (MinIO storage)	KEEP	Works well. Ensure consistent namespace in MinIO.
Sales inbox PDF harvesting	RETHINK	Keep if actively used. Remove hardcoded credentials. Move to background worker with proper scheduling.
Per-user mailbox credential storage	RETHINK	Currently in-memory (lost on restart). Move to encrypted database storage or Redis with proper key management.
Inline HTML email templates (4+ files)	CUT	Replace with proper template system (MJML or html template files with variable substitution).
V1 + V2 email fetch endpoints	CUT	Consolidate into single implementation.

4. Expense / Invoice Features

Feature	Verdict	Rationale
Expense CRUD with receipt upload	KEEP	Core operational feature.
Expense filtering (date, payer, category, payment)	KEEP	Essential for operations team.
Expense CSV export	KEEP	Business reporting need. Fix EUR subtotal + 5% processing fee logic.
Expense PDF export	KEEP	Business reporting need. Fix 2% net10 discount logic.
Expense mark-paid (single + bulk)	KEEP	Essential workflow.
Invoice generation from expenses	RETHINK	Keep functionality. Replace comma-separated expense_ids with proper join table. Add database transaction support.
Invoice payment tracking	KEEP	Core billing workflow.
Invoice PDF generation	RETHINK	Keep. Standardize on single PDF library (@pdfme). Move templates external.
Orphaned invoice cleanup (admin)	CUT	Should be unnecessary with proper relational integrity. Keep a reconciliation check as admin tool.
Currency conversion (Frankfurter API)	KEEP	Needed for multi-currency expenses. Add proper in-memory caching instead of file-based.

5. File Management Features

Feature	Verdict	Rationale
File upload/download via MinIO	KEEP	Works well. Move credentials to env vars.
File browser with folder management	KEEP	Core feature for document access.
File rename/delete	KEEP	Standard file operations.
File preview (presigned URLs)	KEEP	Good pattern.
Email attachments surfaced in file browser	KEEP	Useful cross-reference for sales team.
File audit logging	RETHINK	Currently only prints to console. Wire up to actual audit_logs table.

6. Admin Features

Feature	Verdict	Rationale
Audit log viewer	KEEP	Essential compliance feature. Expand coverage to all domains.
Alert monitoring system (per-type counters, cooldowns)	KEEP	Valuable operational feature. Enhance with dashboard.
Reminder settings management	KEEP	Needed if reminder system is re-enabled.
Alert settings management	KEEP	Operational configuration.
System health dashboard	RETHINK	Currently basic stats. Build proper task execution dashboard.
Webhook event monitor	RETHINK	Currently in-memory only (lost on restart). Persist to database.
Debug endpoints (NocoDB config, OIDC session, connectivity tests)	CUT	Security risk. Replace with admin-only health check endpoint.
Destructive test endpoints (EOI cleanup, berth connection test)	CUT	No place in production. Remove entirely.

7. Auth / Security Features

Feature	Verdict	Rationale
Keycloak OIDC SSO	KEEP	Works well. Clean up implementation (single composable, remove dev bypass from production).
Role-based access (admin/sales/user)	KEEP	Correct model for the organization.
Session caching with TTL + jitter	KEEP	Good performance pattern.
Circuit breaker on Keycloak calls	KEEP	Important resilience pattern.
Dev auth bypass	RETHINK	Move to server-only config. Add build-time assertion preventing production use.
Internal service auth (x-tag header)	CUT	Replace with proper HMAC with rotatable env-var secret.
Feature flag system	RETHINK	The Vuetify→Maritime migration use case is moot. Keep the pattern (dependency chains, rollout %, user targeting) for future feature rollouts.

8. UI / Frontend Features

Feature	Verdict	Rationale
Maritime Design System (glassmorphism tokens)	KEEP	Distinctive visual identity. Reimplement as Tailwind CSS theme extensions.
Vuetify components	CUT	Drop entirely. Maritime-only in rebuild.
Nuxt UI v3 components	RETHINK	Evaluate overlap with Headless UI / Radix Vue. Pick one component library.
Dashboard layout (sidebar, mobile nav)	RETHINK	Rebuild cleanly with single component system. Current version has ~40% duplicate code for dual-system support.
Toast notifications	KEEP	Standard UX pattern.
PWA (workbox, auto-update, 20s sync)	RETHINK	Evaluate if sales team actually needs offline/install. If yes, reduce sync frequency. If no, cut.
Chart.js visualizations	KEEP	Useful for analytics views.

9. Iframe Embeds

Feature	Verdict	Rationale
Metabase analytics dashboard	RETHINK	Keep as iframe OR build native charts. Custom charts integrate better with CRM design language.
NocoDB EOI queue view	CUT	Replace with native CRM page. NocoDB views bypass auth and don't match UI.
NocoDB berth gallery view	CUT	Replace with native CRM berth browser.
Webmail iframe	CUT	Replace with native email UI (EmailCommunication component already exists).
Port Nimara AI	KEEP	Separate service, keep as iframe.
Site analytics (Umami)	KEEP	Read-only, low priority. Keep as iframe.
Social media marketing	KEEP	Third-party service. Keep as iframe.
Client support ticketing	KEEP	Separate system. Keep as iframe.

10. Infrastructure / Architecture

Feature	Verdict	Rationale
NocoDB as primary database	CUT	Replace with PostgreSQL + Drizzle ORM. Gains: transactions, migrations, JOINs, type safety, foreign keys. NocoDB can optionally remain as admin read view.
Single Nuxt container deployment	KEEP	Simple, effective for team size. Add Redis sidecar for sessions/jobs/caching.
Docker + Gitea CI/CD	RETHINK	Keep. Add test stage before image publish.
Nitro experimental tasks + node-cron	CUT	Replace with BullMQ + Redis for proper job tracking, retries, dead letter queues.
In-memory state (webhook store, credential cache, IMAP pool)	CUT	Replace with Redis for shared state that survives restarts.
Console.log-based logging	CUT	Replace with structured logger (consola with levels).
Hardcoded NocoDB table IDs throughout code	CUT	Moot with PostgreSQL migration. If keeping NocoDB: centralize in single config module.

---

Summary Counts

Verdict	Count
KEEP	38
RETHINK	27
CUT	16

The rebuild preserves ~80% of current functionality (keep + rethink) while cutting dead code, security risks, and architectural debt.