Learn about Clawdbot AI assistant streaming responses, chunking, typing indicators, and real-time message delivery.

Clawdbot Streaming Overview

Clawdbot supports streaming responses for real-time AI interactions. Streaming enables typing indicators, partial message updates, and chunked delivery for long responses.

Streaming Modes

Full Streaming

Full streaming delivers tokens as they're generated:

{
  "streaming": {
    "mode": "full",
    "chunkSize": 50
  }
}

Chunked Delivery

Chunked delivery splits long responses:

{
  "streaming": {
    "mode": "chunked",
    "maxLength": 4000,
    "delimiter": "\n\n"
  }
}

Provider Support

Provider	Streaming	Typing	Chunking
WhatsApp	Partial	Yes	Yes
Telegram	Full	Yes	Yes
Discord	Full	Yes	Yes
WeChat	No	No	Yes
Feishu	Partial	Yes	Yes

Typing Indicators

Configuration

{
  "typing": {
    "enabled": true,
    "interval": 3000,
    "stopOnComplete": true
  }
}

Behavior

Indicators sent every interval ms
Automatically stopped on response completion
Provider-specific implementation

Message Chunking

Why Chunking?

Platform message length limits
Better user experience
Reduced timeout risk

Chunk Settings

Setting	Default	Description
`maxLength`	4000	Max characters per chunk
`delimiter`	`\n\n`	Preferred split point
`delay`	500	Ms between chunks

Example

Long response automatically split:

[Chunk 1] First part of response...

[Chunk 2] Continuation...

[Chunk 3] Final part...

Stream Events

Event Types

Event	Description
`stream_start`	Streaming began
`stream_chunk`	New content chunk
`stream_end`	Streaming complete
`stream_error`	Error occurred

WebSocket Events

ws.on('message', (data) => {
  const event = JSON.parse(data);

  switch(event.type) {
    case 'stream_start':
      // Show typing indicator
      break;
    case 'stream_chunk':
      // Append content
      break;
    case 'stream_end':
      // Hide typing indicator
      break;
  }
});

Buffer Management

Configuration

{
  "buffer": {
    "maxSize": 10000,
    "flushInterval": 100,
    "strategy": "adaptive"
  }
}

Strategies

Strategy	Description
`immediate`	Send each token immediately
`batched`	Batch tokens by interval
`adaptive`	Adjust based on network

Error Handling

Retry on Stream Error

{
  "streaming": {
    "retry": {
      "enabled": true,
      "maxAttempts": 3,
      "backoff": "exponential"
    }
  }
}

Fallback to Non-Streaming

If streaming fails, Clawdbot falls back to non-streaming delivery automatically.

Performance Tips

Adjust chunk size based on provider limits
Enable typing indicators for better UX
Use adaptive buffering for variable networks
Monitor stream latency for optimization

Next Steps

Clawdbot Concepts - Core concepts
Clawdbot Channels - Channel setup
Clawdbot Gateway - Gateway configuration

Clawdbot Streaming

On this page