Enhanced Messenger Integration (Telegram Voice, WhatsApp, Channel Abstraction)

[dzianisv]

Overview

Enhance messenger integrations inspired by OpenClaw, a popular personal AI assistant project (167k stars) with comprehensive multi-channel support.

Features

1. Voice Messages in Telegram (High Priority)

• Handle voice notes sent to the Telegram bot
• Use existing Whisper STT server to transcribe audio
• Send transcription to OpenCode, return response as text
• Support opus/ogg audio format conversion

2. WhatsApp via Baileys (High Priority)

• Add WhatsApp channel using Baileys library
• QR code pairing flow for device linking
• Store credentials in ~/.opencode-manager/credentials/whatsapp/
• Per-chat session management (like Telegram)
• Message queueing to prevent race conditions

3. Pairing Security Model (High Priority)

• Replace simple allowlist with pairing code system
• Generate 8-character codes for unknown senders
• CLI command: opencode-manager pairing approve <channel> <code>
• 60-minute TTL for pending requests
• Max 3 pending pairing requests at a time
• Store approved contacts in database

4. Channel Abstraction Layer (Medium Priority)

• Create Channel interface with common capabilities
• Define ChannelCapabilities type (chatTypes, media, threads, reactions)
• Abstract message sending/receiving
• Support for text, audio, images, and files
• Per-channel configuration in settings

5. Telegram Settings UI (Medium Priority)

• Add Telegram settings page in web UI
• Configure bot token via UI (not just env vars)
• Manage allowlist via UI
• View active sessions
• Test bot connectivity

Technical Approach

Channel Interface (TypeScript)

interface Channel {
  id: string;
  name: string;
  capabilities: ChannelCapabilities;
  
  start(): Promise<void>;
  stop(): Promise<void>;
  send(chatId: string, message: OutboundMessage): Promise<void>;
  onMessage(handler: MessageHandler): void;
}

interface ChannelCapabilities {
  chatTypes: ('direct' | 'group' | 'channel')[];
  media: boolean;
  voice: boolean;
  reactions: boolean;
  threads: boolean;
}

Libraries

• Telegram: grammy (already using)
• WhatsApp: @whiskeysockets/baileys
• Audio: ffmpeg for format conversion

Acceptance Criteria

• Send voice note to Telegram bot → gets transcribed and responded to
• Link WhatsApp via QR code → send/receive messages
• Unknown sender gets pairing code → approved via CLI → can chat
• Telegram settings configurable via web UI
• All channels use common Channel interface

References

• OpenClaw channels
• OpenClaw pairing
• Baileys library
• grammy voice handling