Can I vibe code Crisp?

Build a full-stack, single-tenant live chat and customer support application using Node.js (TypeScript), Next.js (App Router), PostgreSQL, Prisma, and Tailwind CSS.

CORE ARCHITECTURE & DATA MODEL:
Create PostgreSQL models for:
- Account / Agent: id, email, passwordHash, name, role (ADMIN, AGENT).
- Conversation: id, channelType (LIVE_CHAT, EMAIL), status (OPEN, PENDING, CLOSED), assignedAgentId, customerId, createdAt, updatedAt.
- Customer: id, email, name, avatarUrl, shopifyCustomerId, metadata (JSONB).
- Message: id, conversationId, senderType (CUSTOMER, AGENT, BOT), senderId, content, attachments (JSONB), createdAt.
- KnowledgeArticle: id, title, slug, content, isPublished, embedding (Vector).

LIVE CHAT WIDGET & REAL-TIME ENGINE:
1. Implement a lightweight client-side JavaScript snippet (`/public/widget.js`) that injects an iframe chat widget into any external HTML page.
2. Establish real-time bidirection communication using WebSockets (Socket.io or WS) between the chat widget, server, and internal agent dashboard.
3. Chat widget must persist session across page navigations using local storage.

SHOPIFY / E-COMMERCE INTEGRATION:
1. Build an API endpoint `/api/integrations/shopify/customer-context` that accepts a customer email or Shopify ID and fetches recent orders, tracking URLs, and lifetime spend via the Shopify Admin GraphQL API.
2. Display this customer context sidebar next to active conversations in the Agent UI.

EMAIL INTEGRATION (INBOUND & OUTBOUND):
1. Create a webhook endpoint `/api/webhooks/inbound-email` (compatible with SendGrid or Postmark inbound parse) that automatically creates or updates a Conversation and records the inbound Message.
2. Integrate outbound email delivery via nodemailer or Postmark SDK so agents can reply to email-based conversations directly from the UI.

AI AUTO-REPLY / KNOWLEDGE BASE RAG:
1. Create a Knowledge Base management UI for creating and editing help articles.
2. Generate vector embeddings for published articles using OpenAI API (`text-embedding-3-small`).
3. Implement an auto-reply trigger: when a customer submits a live chat or email message, perform a vector search against KnowledgeArticle embeddings. If similarity > 0.82, generate a contextual draft using `gpt-4o-mini` and suggest it to the agent, or auto-send if enabled.

AGENT DASHBOARD UI:
1. Build a three-pane layout: Channel/Inbox Nav on the left, Active Conversation thread in center, Customer Data & Shopify Context panel on the right.
2. Include status toggles (Open/Pending/Closed), agent assignment drop-downs, canned response shortcuts (`/` trigger), and private internal notes inside threads.

EXPLICITLY OUT OF SCOPE:
Do not build native mobile apps, WhatsApp/Instagram integrations, or audio/video calling. Focus exclusively on Web Chat Widget, Email handling, Shopify Context Sync, and AI RAG support drafts.

$ each button prefixes agent-specific run instructions · build your own product, never copy proprietary code, trademarks or designs