[Feature Request] Add Streaming Support for Telegram Bot in TrustClaw

[d-dev0101]

Problem

When using TrustClaw via Telegram bot, responses are sent as complete messages rather than streaming incrementally. This creates a poor user experience for longer responses, as users must wait for the entire response to be generated before seeing anything.

Proposed Solution

Implement pseudo-streaming for Telegram bot responses using the editMessageText API. Since Telegram doesn't support true streaming, we can simulate it by:

1. Send an initial "thinking..." message
2. Split the response into chunks (by sentence or character count)
3. Progressively edit the message to append new chunks
4. Create a typing effect that feels responsive

Implementation

I've created a complete implementation ready for integration:

Download: https://backend.composio.dev/api/v3/sl/3719VB69l8

The package includes:

• telegram_streaming_complete.py - Full Python implementation
• INTEGRATION_GUIDE.md - Step-by-step integration guide

Key features:

• Configurable chunking (by sentence or character count)
• Rate-limit handling (Telegram API limits: ~3 edits/second)
• Fallback to regular sending if streaming fails
• Minimal code changes required

Benefits

✅ Better UX - Users see responses appear incrementally
✅ Perceived performance - Feels faster even if generation time is same
✅ User feedback - "thinking..." indicator shows the bot is working
✅ Production-ready - Handles rate limits and errors gracefully

Integration Example

# Before:
bot.send_message(chat_id, response_text)

# After:
streamer = TelegramStreamingHelper(bot_token)
streamer.send_streaming_response(
    chat_id=chat_id,
    full_text=response_text,
    method="sentence",  # Split by sentences
    delay=0.2  # 200ms between updates
)

Context

This was requested by a TrustClaw user who wants streaming responses in Telegram conversations. The implementation is tested and ready to use.

Let me know if you need any clarifications or modifications!

[Uday-sidagana]

Hey @d-dev0101, we've responded on your Issue #2812