agentnexus

AgentNexus: Python Library for AI/LLM Agent Development with UI-Driven Workflows

⚠️ ALPHA STAGE: This library is currently in alpha.

🚀 Overview

AgentNexus simplifies the creation of LLM and AI agents with interactive user interfaces and structured workflows. It automatically generates agent manifests to enable seamless discovery, distributed deployment, and cross-agent communication.

The library provides a declarative approach to building AI agents using Python decorators, allowing developers to focus on agent logic rather than infrastructure.

🌟 Key Features

🏗️ Advanced Agent Architecture

• Declarative Development: Create complex agents using intuitive Python decorators
• Multi-Action Support: Define multiple agent actions within a single agent
• Workflow-Driven Design: Build multi-step UI workflows with state management

🖥️ Event-Driven UI System

• Rich Component Library: Tables, forms, code editors, and markdown displays
• Automatic Event Handling: Simplified component event management
• Context-Aware State: Preserve state across workflow steps and sessions

🔍 Comprehensive Manifest Generation

• Standard Protocol: JSON schema for agents, actions, workflows, and UI components
• Cross-Platform Compatibility: Works with any platform supporting the manifest spec
• Distributed Architecture: Enables decentralized agent discovery and execution

📸 Screenshots

🛠️ Installation

Installing from PyPI (Recommended)

# Install the latest release
pip install agentnexus

# Or with a specific version
pip install agentnexus==0.1.0

Installing with uv (Faster Installation)

pip install uv  # If you don't have uv installed
uv pip install agentnexus

🔧 Local Development

Prerequisites

• Python 3.8+
• Git

Install from Source

# Clone the repository
git clone https://github.com/MuhammadYossry/agentnexus.git
cd agentnexus

# Option 1: Using uv (recommended for faster dependency resolution)
pip install uv  # If you don't have uv installed
uv venv
uv pip install -r requirements.txt

# Option 2: Using pip and venv
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate
pip install -r requirements.txt

Running the Application

# Using uv
uv run uvicorn main:app --port 9200 --reload

# Using regular Python (if installed with venv)
uvicorn main:app --port 9200 --reload

The server will start at http://localhost:9200, and the auto-generated manifest will be available at http://localhost:9200/agents.json.

Running Tests

# Using uv
uv run pytest

# Or using the included script
python run_tests.py

🚀 Quick Start

Creating a Basic Agent

from agentnexus.base_types import AgentConfig, Capability, ActionType
from agentnexus.action_manager import agent_action

# Define agent
my_agent = AgentConfig(
    name="Simple Assistant",
    version="1.0.0",
    description="Basic AI assistant with simple capabilities",
    capabilities=[
        Capability(
            skill_path=["Utilities", "Text"],
            metadata={"features": ["Summarization", "Translation"]}
        )
    ]
)

# Create an action
@agent_action(
    agent_config=my_agent,
    action_type=ActionType.GENERATE,
    name="Summarize Text",
    description="Creates a concise summary of longer text"
)
async def summarize_text(input_data):
    """Summarize the provided text."""
    text = input_data.text
    max_length = getattr(input_data, "max_length", 100)

    # Summarization logic would go here
    summary = text[:max_length] + "..."  # Simplified example

    return {"summary": summary, "original_length": len(text)}

Setting Up FastAPI Integration

from fastapi import FastAPI
from agentnexus.manifest_generator import AgentManager

app = FastAPI()
agent_manager = AgentManager(base_url="http://localhost:8000")
agent_manager.add_agent(my_agent)
agent_manager.setup_agents(app)

# Run with: uvicorn main:app --reload

📚 Key Concepts

• Agent Config: Defines an agent's metadata and capabilities
• Actions: Standalone agent capabilities, triggered via API endpoints
• Workflows: Multi-step processes with state management
• UI Components: Interactive UI elements with event handling
• Context Management: State preservation across workflow steps
• Manifest: Standardized JSON schema for agent discovery and integration

🌐 Auto-Generated Endpoints

The library automatically creates these endpoints:

• GET /agents.json - List all available agents
• GET /agents/{agent_slug}.json - Get detailed manifest for a specific agent
• POST /agents/{agent_slug}/actions/{action_name} - Trigger agent actions
• POST /agents/{agent_slug}/workflows/{workflow_id}/steps/{step_id} - Execute workflow steps

🧩 UI Components

AgentNexus includes several built-in UI components:

• FormComponent: Interactive forms for data collection
• TableComponent: Display and interact with tabular data
• CodeEditorComponent: Code editing with syntax highlighting
• MarkdownComponent: Formatted text display

📖 Documentation

Comprehensive documentation is under development. For now, please refer to the examples in the agents directory.

🛣️ Project Status

AgentNexus is in alpha stage and under active development. The API may change significantly before the first stable release. Current limitations include:

• Limited production testing
• Incomplete documentation
• Some API instability
• Limited example collection

🤝 Contributing

Contributions are welcome! Help us improve AgentNexus by:

1. Reporting bugs
2. Suggesting features
3. Submitting pull requests
4. Improving documentation

📜 License

MIT License

🌟 Star Us

If you find AgentNexus useful, please give us a star on GitHub!