diff --git a/.dev.vars.example b/.dev.vars.example
index b97e98a..6536148 100644
--- a/.dev.vars.example
+++ b/.dev.vars.example
@@ -4,7 +4,9 @@ NEXTJS_ENV=development
 # Drizzle Kit credentials for D1
 # Get your Account ID from: https://dash.cloudflare.com/ (right sidebar)
 # Get your API Token from: https://dash.cloudflare.com/profile/api-tokens
+# Get your Database ID from: wrangler d1 list (or create with: wrangler d1 create <name>)
 CLOUDFLARE_ACCOUNT_ID=your-account-id-here
+CLOUDFLARE_D1_DATABASE_ID=your-database-id-here
 CLOUDFLARE_D1_TOKEN=your-api-token-here
 CLOUDFLARE_R2_URL=your-r2-url-here
 CLOUDFLARE_API_TOKEN=your-api-token-here
diff --git a/PROJECT_BRIEF.md b/PROJECT_BRIEF.md
new file mode 100644
index 0000000..4f9714a
--- /dev/null
+++ b/PROJECT_BRIEF.md
@@ -0,0 +1,310 @@
+# Project Brief: Fullstack Next.js + Cloudflare CRM
+
+**Created**: 2025-11-08
+**Status**: Ready for Planning
+**Purpose**: Learning exercise to understand Next.js 15 + Cloudflare Workers integration
+
+---
+
+## Vision
+
+A lightweight CRM built on the fullstack-next-cloudflare template to learn modern fullstack patterns: Next.js App Router, Cloudflare D1/R2, Server Actions, and module-sliced architecture.
+
+---
+
+## Problem/Opportunity
+
+**Learning objective**: Understand how to build production-grade features on Cloudflare's edge platform by implementing real-world CRM functionality.
+
+**Why CRM as the learning vehicle**:
+- Multi-entity relationships (contacts ↔ deals)
+- CRUD operations with validation
+- Data modeling with foreign keys
+- UI patterns (lists, forms, boards)
+- Follows existing architecture (todos module)
+
+---
+
+## Target Audience
+
+- **Primary user**: You (Jez) - exploring the stack
+- **Scale**: Single user for learning (no multi-tenancy needed)
+- **Context**: Educational project, not production SaaS
+- **Data**: Can use synthetic/test data
+
+---
+
+## Core Functionality (MVP)
+
+### 1. Contacts Module
+**Essential**:
+- ✅ Create, read, update, delete contacts
+- ✅ Fields: firstName, lastName, email, phone, company, jobTitle, notes
+- ✅ Search/filter by name, email, company
+- ✅ Tag system (many-to-many: contacts ↔ tags)
+- ✅ User-specific tags with colors
+
+**Deferred to Phase 2** (keep MVP lean):
+- ❌ Activity timeline (calls, meetings, notes)
+- ❌ Avatar uploads to R2
+- ❌ Email integration
+- ❌ Import/export
+
+### 2. Deals/Pipeline Module
+**Essential**:
+- ✅ Create, read, update, delete deals
+- ✅ Fields: title, value, currency, stage, expectedCloseDate, description
+- ✅ Link deal to contact (simple 1:1 relationship)
+- ✅ Pipeline board view (simple columns by stage)
+- ✅ Fixed stages: Prospecting → Qualification → Proposal → Negotiation → Closed Won/Lost
+
+**Deferred to Phase 2**:
+- ❌ Custom user-defined stages
+- ❌ Drag-and-drop stage changes
+- ❌ Deal probability/forecasting
+- ❌ Multiple contacts per deal
+
+### 3. Dashboard Integration
+**Essential**:
+- ✅ Add navigation to /dashboard/contacts and /dashboard/deals
+- ✅ Simple metrics cards (total contacts, active deals, pipeline value)
+
+**Deferred**:
+- ❌ Charts/graphs
+- ❌ Activity feed
+- ❌ Advanced analytics
+
+---
+
+## Tech Stack (Validated)
+
+Uses existing template stack - no changes needed:
+
+- **Frontend**: Next.js 15.4.6 (App Router) + React 19 + TypeScript
+- **UI**: Tailwind v4 + shadcn/ui + Lucide icons
+- **Backend**: Cloudflare Workers with Static Assets (@opennextjs/cloudflare)
+- **Database**: Cloudflare D1 (SQLite) with Drizzle ORM
+- **Storage**: Cloudflare R2 (not using for MVP - deferred avatars)
+- **Auth**: Better Auth (already configured)
+- **Forms**: React Hook Form + Zod validation
+- **Deployment**: Cloudflare Workers (via GitHub Actions)
+
+**Why this stack works for learning**:
+- ✅ Modern patterns (Server Actions, RSC)
+- ✅ Edge-first architecture
+- ✅ Type-safe end-to-end (TypeScript + Drizzle + Zod)
+- ✅ Template already has auth, DB, migrations configured
+- ✅ Follows module-sliced pattern (easy to extend)
+
+---
+
+## Research Findings
+
+### Existing Template Analysis
+
+**What's already built** (from /home/jez/Documents/fullstack-next-cloudflare-demo):
+- ✅ **Auth module**: Better Auth with email/password + Google OAuth
+- ✅ **Todos module**: Complete CRUD example with categories, priorities, status
+- ✅ **Database setup**: D1 + Drizzle + migrations working
+- ✅ **R2 integration**: Image upload pattern (in todos for cover images)
+- ✅ **Module architecture**: `src/modules/[feature]/` with actions/, components/, schemas/
+- ✅ **UI components**: 13 shadcn/ui components configured
+- ✅ **Deployment**: GitHub Actions workflow ready
+
+**Pattern to follow**:
+The `src/modules/todos/` structure is the perfect blueprint:
+```
+todos/
+├── actions/           # Server actions (create, get, update, delete)
+├── components/        # UI components (form, card, list)
+├── models/            # Enums and types
+└── schemas/           # Drizzle + Zod schemas
+```
+
+We'll replicate this for `contacts/` and `deals/` modules.
+
+### Technical Validation
+
+**✅ D1 Relational Data**:
+- Drizzle ORM supports foreign keys and joins
+- Template already has `todos → categories` relationship
+- Contacts ↔ Deals will work the same way
+
+**✅ Many-to-Many Tags**:
+- Need junction table: `contacts_to_tags`
+- Drizzle example in their docs: https://orm.drizzle.team/docs/rqb#many-to-many
+
+**✅ Server Actions Performance**:
+- Template uses server actions for all mutations
+- Edge runtime = fast globally
+- No API route boilerplate needed
+
+**Known Challenges**:
+1. **Junction table queries** - Drizzle syntax for many-to-many can be verbose
+   - Mitigation: Study existing `todos.categoryId` pattern, extend to junction table
+2. **Pipeline board UI** - Kanban layout without drag-drop library
+   - Mitigation: Simple CSS Grid columns, manual stage update dropdown (defer drag-drop to Phase 2)
+3. **Search implementation** - D1 doesn't have full-text search
+   - Mitigation: Use SQL `LIKE` queries for MVP (good enough for learning)
+
+---
+
+## Scope Validation
+
+### Why Build This?
+**Learning objectives met**:
+- ✅ Practice module-sliced architecture
+- ✅ Understand Drizzle ORM relationships (1:1, many-to-many)
+- ✅ Learn Server Actions data mutation patterns
+- ✅ Explore D1 migrations workflow
+- ✅ Build complex forms with validation
+- ✅ Create dashboard visualizations
+- ✅ Deploy to Cloudflare edge
+
+**Why NOT use existing CRM**:
+- This is about learning the stack, not production use
+- Building from scratch teaches architectural patterns
+- Template provides 80% foundation (auth, DB, UI), we add 20% (domain logic)
+
+### Why This Scope?
+**MVP is deliberately minimal** to focus on learning core patterns:
+- 2 main entities (contacts, deals) = practice relationships
+- Tags system = practice many-to-many
+- Pipeline board = practice UI state management
+- Dashboard metrics = practice aggregations
+
+**Deferred features** prevent scope creep:
+- Activity logging (complex timeline UI)
+- Avatars (R2 already demonstrated in todos)
+- Custom stages (adds complexity)
+- Advanced analytics (not core learning)
+
+**Time investment** = ~6-8 hours (~6-8 minutes with Claude Code)
+- Realistic for learning project
+- Can complete in 1-2 sessions
+- Leaves room for experimentation
+
+### What Could Go Wrong?
+
+**Risk 1: Overcomplicating relationships**
+- *What*: Trying to add too many foreign keys (deals → contacts → companies → industries...)
+- *Mitigation*: Stick to MVP scope (contacts ↔ tags, deals → contacts). No nested hierarchies.
+
+**Risk 2: UI perfectionism**
+- *What*: Spending hours on drag-and-drop Kanban, animations, etc.
+- *Mitigation*: Use simple table/grid layouts. Focus on functionality, not polish.
+
+**Risk 3: Scope creep during build**
+- *What*: "While I'm here, let me add email integration..."
+- *Mitigation*: Strict adherence to MVP checklist. Document ideas for Phase 2.
+
+---
+
+## Estimated Effort
+
+**Total MVP**: ~6-8 hours (~6-8 minutes human time with Claude Code)
+
+**Breakdown**:
+- Setup (clone, configure D1, run migrations): 30 min
+- Contacts module (schema, actions, UI, tags): 2.5 hours
+- Deals module (schema, actions, UI, board): 2 hours
+- Dashboard integration (nav, metrics): 1 hour
+- Testing & seed data: 1 hour
+- Documentation: 30 min
+
+**Phase 2** (optional extensions):
+- Activity timeline: +2 hours
+- Avatar uploads: +1 hour
+- Drag-drop Kanban: +2 hours
+- Custom stages: +1.5 hours
+- Advanced search: +2 hours
+
+---
+
+## Success Criteria (MVP)
+
+**Functional Requirements**:
+- [ ] Can create, edit, delete, search contacts
+- [ ] Can assign multiple tags to contacts
+- [ ] Can create tags with colors
+- [ ] Can create, edit, delete deals
+- [ ] Deals link to contacts (dropdown selector)
+- [ ] Pipeline board shows deals in columns by stage
+- [ ] Dashboard shows: total contacts, active deals, pipeline value
+- [ ] All data isolated to logged-in user
+- [ ] Forms have proper validation (Zod schemas)
+- [ ] UI responsive on mobile/desktop
+
+**Technical Requirements**:
+- [ ] Follows module-sliced architecture (`src/modules/contacts/`, `src/modules/deals/`)
+- [ ] Uses Server Actions (not API routes)
+- [ ] Database migrations run successfully (local + production)
+- [ ] Type-safe end-to-end (TypeScript + Drizzle + Zod)
+- [ ] shadcn/ui components used consistently
+- [ ] Deploys to Cloudflare Workers without errors
+
+**Learning Objectives**:
+- [ ] Understand how to structure multi-entity features
+- [ ] Practice Drizzle ORM relationships (foreign keys, joins, many-to-many)
+- [ ] Learn Server Actions patterns for CRUD
+- [ ] Experience D1 migrations workflow
+- [ ] Build complex forms with React Hook Form + Zod
+
+---
+
+## Next Steps
+
+### If Proceeding (Recommended)
+
+1. **Exit plan mode** and start implementation
+2. **Clone project** to `/home/jez/Documents/fullstack-next-cloudflare-crm`
+3. **Configure local Cloudflare**:
+   - Create new D1 database: `npx wrangler d1 create fullstack-crm`
+   - Update `wrangler.jsonc` with new database ID
+   - Set up `.dev.vars` with Better Auth secrets
+4. **Implement in phases**:
+   - Phase 1: Project setup + database schema
+   - Phase 2: Contacts module
+   - Phase 3: Deals module
+   - Phase 4: Dashboard integration
+   - Phase 5: Testing & documentation
+5. **Deploy when ready** (Cloudflare account setup)
+
+### If Refining Scope
+
+**Want simpler?**
+- Skip tags (just contacts + deals)
+- Skip pipeline board (simple table view)
+- Reduces to ~4 hours
+
+**Want more ambitious?**
+- Add activity timeline
+- Add R2 avatar uploads
+- Add custom stages
+- Increases to ~10-12 hours
+
+---
+
+## Research References
+
+- **Template repo**: https://github.com/jezweb/fullstack-next-cloudflare (forked from ifindev)
+- **Local codebase**: /home/jez/Documents/fullstack-next-cloudflare-demo
+- **Drizzle ORM relationships**: https://orm.drizzle.team/docs/rqb
+- **shadcn/ui components**: https://ui.shadcn.com/docs
+- **Cloudflare D1 docs**: via `mcp__cloudflare-docs__search_cloudflare_documentation`
+- **Relevant skills**: `~/.claude/skills/cloudflare-d1`, `~/.claude/skills/drizzle-orm-d1`
+
+---
+
+## Recommendation
+
+✅ **Proceed with MVP implementation**
+
+**Rationale**:
+1. Scope is well-defined and realistic (6-8 hours)
+2. Template provides solid foundation (80% already built)
+3. Learning objectives are clear and achievable
+4. No technical blockers (all patterns exist in template)
+5. Can defer advanced features to Phase 2 without compromising learning
+
+**This is an excellent learning project** - complex enough to teach real patterns, simple enough to complete without frustration.
diff --git a/README.md b/README.md
index 333f7ab..0feb729 100644
--- a/README.md
+++ b/README.md
@@ -149,6 +149,7 @@ Edit `.dev.vars` with your credentials:
 ```bash
 # Cloudflare Configuration
 CLOUDFLARE_ACCOUNT_ID=your-account-id
+CLOUDFLARE_D1_DATABASE_ID=your-database-id
 CLOUDFLARE_D1_TOKEN=your-api-token
 
 # Authentication Secrets
@@ -263,6 +264,10 @@ wrangler d1 create your-app-name
 # Output will show:
 # database_name = "your-app-name"
 # database_id = "xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx"
+#
+# Copy the database_id - you'll need it for:
+# - wrangler.jsonc (d1_databases.database_id)
+# - .dev.vars (CLOUDFLARE_D1_DATABASE_ID)
 ```
 
 **Create R2 Bucket:**
@@ -329,6 +334,7 @@ openssl rand -base64 32
 ```bash
 # .dev.vars for local development
 CLOUDFLARE_ACCOUNT_ID=your-account-id
+CLOUDFLARE_D1_DATABASE_ID=your-database-id
 CLOUDFLARE_D1_TOKEN=your-api-token
 BETTER_AUTH_SECRET=your-generated-secret
 GOOGLE_CLIENT_ID=your-google-client-id.apps.googleusercontent.com
@@ -403,6 +409,7 @@ Go to your GitHub repository → Settings → Secrets and add:
 
 - `CLOUDFLARE_API_TOKEN` - Your API token from Step 2
 - `CLOUDFLARE_ACCOUNT_ID` - Your account ID
+- `CLOUDFLARE_D1_DATABASE_ID` - Your D1 database ID (from `wrangler d1 create` output)
 - `BETTER_AUTH_SECRET` - Your auth secret
 - `GOOGLE_CLIENT_ID` - Your Google client ID
 - `GOOGLE_CLIENT_SECRET` - Your Google client secret
diff --git a/drizzle.config.ts b/drizzle.config.ts
index b3c0c64..30fb1aa 100644
--- a/drizzle.config.ts
+++ b/drizzle.config.ts
@@ -13,7 +13,7 @@ export default defineConfig({
     driver: "d1-http",
     dbCredentials: {
         accountId: process.env.CLOUDFLARE_ACCOUNT_ID!,
-        databaseId: "757a32d1-5779-4f09-bcf3-b268013395d4",
+        databaseId: process.env.CLOUDFLARE_D1_DATABASE_ID!,
         token: process.env.CLOUDFLARE_D1_TOKEN!,
     },
 });