Skip to main content

​
Streaming

PraisonAI TypeScript SDK supports streaming responses, allowing you to receive agent output in real-time as it’s generated.

​
Quickstart

import { Agent } from 'praisonai';

const agent = new Agent({
  instructions: "You are a helpful assistant",
  stream: true  // Enable streaming (default)
});

const response = await agent.chat("Tell me a story");
// Response streams to console in real-time

​
Configuration

​
Enable Streaming

Streaming is enabled by default: 
const agent = new Agent({
  instructions: "You are helpful",
  stream: true  // Default value
});

​
Disable Streaming

For batch processing or when you need the complete response: 
const agent = new Agent({
  instructions: "You are helpful",
  stream: false
});

const response = await agent.chat("Hello");
// Returns complete response after generation finishes

​
Streaming Behavior

​
With Verbose Mode

When both stream and verbose are enabled: 
const agent = new Agent({
  instructions: "You are helpful",
  stream: true,
  verbose: true
});

await agent.chat("Explain AI");
// Output streams to console with formatting

​
Silent Streaming

Stream without console output: 
const agent = new Agent({
  instructions: "You are helpful",
  stream: true,
  verbose: false
});

const response = await agent.chat("Hello");
// Streams internally, returns complete response

​
Use Cases

​
Interactive Chat

const agent = new Agent({
  instructions: "You are a conversational assistant",
  stream: true,
  verbose: true
});

// User sees response as it's generated
await agent.chat("Tell me about space exploration");

​
Long-Form Content

const writer = new Agent({
  instructions: "You are a creative writer",
  stream: true
});

// Stream a long story
await writer.chat("Write a 1000-word story about a robot");

​
Code Generation

const coder = new Agent({
  instructions: "You are a code assistant",
  stream: true
});

// Stream code as it's written
await coder.chat("Write a React component for a todo list");

​
Streaming with Tools

Tool calls are handled automatically during streaming: 
const getWeather = (city: string) => `Weather in ${city}: 22°C`;

const agent = new Agent({
  instructions: "You are a weather assistant",
  stream: true,
  tools: [getWeather]
});

await agent.chat("What's the weather in Paris?");
// Tool is called, then response streams

​
Performance Considerations

​
When to Use Streaming

  • Interactive applications: Real-time user feedback
  • Long responses: Show progress during generation
  • Chat interfaces: Natural conversation flow

​
When to Disable Streaming

  • Batch processing: Processing many requests
  • API endpoints: Need complete response for JSON
  • Testing: Easier to validate complete responses
// Batch processing example
const agent = new Agent({
  instructions: "Summarize text",
  stream: false  // Faster for batch
});

const summaries = await Promise.all(
  documents.map(doc => agent.chat(`Summarize: ${doc}`))
);

​
Environment Variables

# Control default streaming behavior
PRAISON_STREAM=true

# Control verbose output
PRAISON_VERBOSE=true

​
Example: Streaming Chat Application

import { Agent } from 'praisonai';
import * as readline from 'readline';

const agent = new Agent({
  instructions: "You are a helpful assistant",
  stream: true,
  verbose: true
});

const rl = readline.createInterface({
  input: process.stdin,
  output: process.stdout
});

async function chat() {
  rl.question('You: ', async (input) => {
    if (input.toLowerCase() === 'exit') {
      rl.close();
      return;
    }
    
    console.log('Assistant: ');
    await agent.chat(input);
    console.log('\n');
    
    chat();
  });
}

chat();

​
See Also