AutoAgents

Auto-generate agents from topics

AutoAgents

AutoAgents automatically generates agent configurations from a topic description.

Installation

npm install praisonai

Quick Start

import { createAutoAgents } from 'praisonai';

const auto = createAutoAgentTeam({
  llm: 'openai/gpt-4o-mini'
});

const result = await auto.generate('Build a web scraper');
console.log(result.agents);
console.log(result.tasks);

Configuration

interface AutoAgentsConfig {
  llm?: string;           // LLM model to use
  pattern?: string;       // Agent pattern
  verbose?: boolean;      // Enable verbose logging
}

Patterns

Pattern	Description
sequential	Agents execute in sequence
parallel	Agents execute in parallel
routing	Route to specific agents

Example

import { createAutoAgents } from 'praisonai';

const auto = createAutoAgentTeam({
  llm: 'openai/gpt-4o-mini',
  pattern: 'sequential',
  verbose: true
});

const result = await auto.generate('Create a data analysis pipeline');

// Result contains:
// - agents: Array of agent configurations
// - tasks: Array of task definitions
// - pattern: The pattern used

CLI Usage

praisonai-ts auto "Build a web scraper"
praisonai-ts auto "Create a data pipeline" --pattern sequential --json

Response Format

interface AutoAgentsResult {
  agents: Array<{
    name: string;
    role: string;
    goal: string;
    instructions: string;
    tools: string[];
  }>;
  tasks: Array<{
    description: string;
    expectedOutput: string;
    agent: string;
  }>;
  pattern: string;
}