feat: Telegram group chat support — agents as community managers (TGRAM-GROUP)

[vybe]

Summary

Enable agents to participate in Telegram group chats, supporting use cases like community management, customer support groups, and multi-user channels. Agents should respond to @mentions, reply in threads, and handle group-specific behaviors (join/leave events, pinned messages, admin commands).

Motivation

Telegram groups are the primary channel for Web3 communities, customer support, and team coordination. Agents acting as community managers need to handle group dynamics: multiple users talking, @mention-based activation, topic threads, and moderation actions.

Requirements

Group Chat Message Handling

• Agent responds when @mentioned by bot username
• Agent responds to direct replies to its own messages
• Configurable: respond to all messages vs @mention-only (per binding)
• Thread/topic support — keep conversation context within a thread

Group-Specific Features

• Handle new member join events (welcome messages, onboarding)
• Handle member leave events (if configured)
• Pin important messages (if bot has admin rights)
• Respect group admin commands (e.g., /mute, /pause to silence the agent)

Multi-Agent in Groups

• Multiple agents can be bound to the same group with different trigger rules
• Route by @mention to specific agent
• Default agent for unaddressed messages (optional, off by default)

Agent Identity

• Bot displays agent's name (set via BotFather per bot, or via display name if using a single bot with multiple agents)
• Agent avatar used as bot profile picture where possible

Rate Limiting & Safety

• Per-group rate limiting to prevent flooding
• Cooldown between responses in active groups
• Max message length handling (split long responses)
• Ignore messages from other bots (prevent bot loops)

Acceptance Criteria

• Agent responds to @mentions in group chats
• Agent responds to direct replies to its messages
• Configurable trigger mode: @mention-only vs all messages
• New member welcome messages work
• Multiple agents can coexist in one group
• Rate limiting prevents message flooding
• Bot loop prevention (ignores other bot messages)
• UI shows group bindings with trigger configuration