usualdork/aiwhisper

Purpose and Entry Points

AIWhisper is a Next.js WhatsApp automation console for configuring AI agents, connecting a WhatsApp Web account, monitoring inbox activity, and generating Gemini-based replies. The browser application begins at src/app/page.tsx, with authenticated/console-style pages grouped under src/app/(app)/layout.tsx. HTTP integration boundaries are Next.js route handlers under src/app/api.

Management Console

The app router serves dashboard, inbox, agents, knowledge-base, activity, and settings views. Feature-level client components encapsulate the operational UI:

• Agent design/listing: src/app/(app)/agents/page.tsx, src/components/agent-designer.tsx, src/components/agent-list.tsx

• Inbox and conversation handling: src/app/(app)/inbox/page.tsx, src/components/inbox-layout.tsx

• Knowledge and WhatsApp QR UX: src/app/(app)/knowledge-base/page.tsx, src/components/knowledge-base-manager.tsx, src/components/qr-login.tsx

API and Application Boundary

Route handlers expose REST-style endpoints consumed by the UI and external callers. They partition the domain into agents, inbox/messages, knowledge ingestion, activity logs, dashboard statistics, and WhatsApp lifecycle:

• src/app/api/agents/route.ts, src/app/api/inbox/messages/[chatId]/route.ts

• src/app/api/knowledge/parse/route.ts, src/app/api/dashboard/stats/route.ts

• src/app/api/whatsapp/init/route.ts, src/app/api/whatsapp/send/route.ts, src/app/api/whatsapp/status/route.ts

AI Response Pipeline

Genkit hosts the AI runtime and defines an intelligent-response-selection flow, backed by Google Gemini credentials supplied through environment configuration. The intended flow is inbound WhatsApp message/context → response-selection flow → generated reply → WhatsApp send operation.

• src/ai/genkit.ts

• src/ai/flows/intelligent-response-selection.ts

• src/lib/ai.ts

src/ai/dev.ts is the separate Genkit development entry point, run alongside Next.js via genkit:dev.

WhatsApp Integration

Baileys provides the persistent WhatsApp Web/WebSocket connection, QR authentication, outbound messaging, status inspection, and logout. The client adapter isolates this stateful external integration from HTTP endpoints.

• src/lib/whatsapp-client.ts

• src/app/api/whatsapp/init/route.ts

• src/app/api/whatsapp/logout/route.ts

This is a material runtime boundary: the Next.js server/API layer invokes a Node-compatible, connection-owning WhatsApp client rather than directly exposing Baileys to the browser.

Persistence and Operational Data

The repository currently contains JSON datasets for agents, conversations, messages, knowledge, logs, and aggregate stats. The database abstraction and server actions are the likely access boundary between routes/UI and persisted state.

• src/lib/db.ts, src/lib/actions.ts

• src/data/agents.json, src/data/conversations.json, src/data/messages.json

• src/data/knowledge.json, src/data/logs.json, src/data/stats.json

Runtime and Dependencies

Primary runtime is Node.js 18+ with Next.js 15 and React. Tailwind/Radix-based UI is rendered by the App Router; Recharts supports analytics views. External services are Google Gemini and WhatsApp Web through Baileys. WhatsApp session state is file-based per the project design, while business/analytics data is represented locally in JSON.