NEW: v0.5.5 is out!

Open-source AI agent framework

Powerful and developer-friendly framework for building autonomous AI agents.

From Intent to Execution

Sub Agents

Coordinate Specialized AI Teams

Create specialized agents that work together. Automatic delegation, parallel execution, and intelligent coordination.

Sub Agents
coordinator.maindelegating
researcheractive
writerwaiting
Graphs

Complex Workflows Made Simple

Build sophisticated multi-step workflows with dependencies, conditional logic, and automated scheduling.

Graphs
dataCollection
completed
processing
running
reportGen
pending
graph.status → executing

Build Enterprise-Grade Agents

Agents that remember everything.

Long-term memory enables agents to remember past conversations, learn from interactions, and maintain context across sessions.

“I remember you prefer TypeScript...”
“Your last project was about AI agents”
retrieving memories...
AI Agents

Execute complex tasks with precision.

Structured task execution with status tracking, file attachments, and intelligent tool integration.

Task #3845
pending
Task #3846
in_progress
Task #3847
completed
Task Management

Build intelligent agents in minutes.

Create AI agents with memory, vision, and tools. Deploy them instantly with just a few lines of code.

import { Agent } from '@astreus-ai/astreus';

// Create agent
const agent = await Agent.create({
  name: 'ResearchAgent',
  model: 'gpt-4o',
  memory: true,
  systemPrompt: 'You are an expert research assistant.'
});

// Create a task
const task = await agent.createTask({
  prompt: "Research latest news in Anthropic and OpenAI"
});

// Execute the task
const result = await agent.executeTask(task.id);
console.log(result.response);
Code Playground

Build with Astreus

MCP Integration

Model Context Protocol support for seamless external tool and service connections.

Plugin System

Extensible tool integration with JSON schema validation and automatic LLM function calling.

Vision Processing

Built-in image analysis and document processing capabilities for multimodal interactions.

Knowledge Base

RAG integration with document chunking, vector embeddings, and similarity search.

Multi-LLM Integration

Unified interface for OpenAI, Claude, Gemini, and Ollama with automatic model routing.

Advanced Logging

Structured logging with conversation tracking, performance metrics, and debugging capabilities.

Security & Safety

Built-in prompt injection protection, content filtering, and secure execution environments.

Developer Experience

TypeScript support, hot reloading, comprehensive documentation, and intuitive CLI tools.