An open-source implementation of Anthropic's multi-agent deep research system as defined in: https://www.anthropic.com/engineering/built-multi-agent-research-system built with the swarms framework

Advanced Research System (Based on Anthropic's Paper)

An enhanced implementation of the orchestrator-worker pattern from Anthropic's paper, "How we built our multi-agent research system," using the swarms framework. This system achieves 90.2% performance improvement over single-agent systems through advanced parallel execution, LLM-as-judge evaluation, and professional report generation with export capabilities.

✨ Key Features

🏗️ Architecture

The system follows a dynamic, multi-phase workflow with enhanced coordination:

                [User Query + Export Options]
                            │
                            ▼
           ┌─────────────────────────────────┐
           │    LeadResearcherAgent          │ (Enhanced Orchestrator)
           │  - Query Analysis & Planning    │
           │  - LLM-as-Judge Evaluation      │
           │  - Iterative Strategy Refinement│
           └─────────────────────────────────┘
                            │ 1. Analyze & Decompose (with thinking process)
                            ▼
       ┌─────────────────────────────────────────┐
       │         Parallel Sub-Tasks              │
       │      (Up to 5 concurrent tasks)         │
       └─────────────────────────────────────────┘
          │           │           │           │
          ▼           ▼           ▼           ▼
    ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐
    │SubAgent 1│ │SubAgent 2│ │SubAgent 3│ │SubAgent N│ (Specialized Workers)
    │Multi-loop│ │Multi-loop│ │Multi-loop│ │Multi-loop│
    │ Search   │ │ Search   │ │ Search   │ │ Search   │
    └──────────┘ └──────────┘ └──────────┘ └──────────┘
          │           │           │           │
          ▼           ▼           ▼           ▼
       ┌─────────────────────────────────────────┐
       │     Enhanced Results Aggregation        │
       │  - Quality Assessment & Confidence      │
       │  - Source Deduplication & Scoring       │
       └─────────────────────────────────────────┘
                            │ 2. Synthesis & LLM-as-Judge Evaluation
                            ▼
           ┌─────────────────────────────────┐
           │    LeadResearcherAgent          │
           │  - Completeness Assessment      │
           │  - Gap Identification           │
           │  - Iterative Refinement         │
           └─────────────────────────────────┘
                            │ 3. Generate Final Report
                            ▼
           ┌─────────────────────────────────┐
           │      Enhanced CitationAgent     │ (Post-Processor)
           │  - Smart Source Descriptions    │
           │  - Professional Citations       │
           │  - Quality Assurance            │
           └─────────────────────────────────┘
                            │ 4. Export & Delivery
                            ▼
              [Final Cited Report + Optional Export]

🔄 Enhanced Workflow Process

1. Strategic Planning: Advanced query analysis with explicit thinking processes and complexity assessment
2. Parallel Research: Multiple ResearchSubagent workers with 3-loop search strategies execute concurrently
3. LLM-as-Judge Evaluation: Sophisticated progress assessment identifies gaps and determines iteration needs
4. Professional Citation: Enhanced processing with intelligent source descriptions and quality indicators
5. Export & Delivery: Optional file export with customizable paths and comprehensive metadata

📦 Installation

Prerequisites

• Python 3.10 or higher
• API keys for Claude (Anthropic) and Exa search

Install with uv (Recommended)

uv provides the fastest and most reliable package management experience:

BASH
# Install uv if you haven't already
curl -LsSf https://astral.sh/uv/install.sh | sh

# Install the package
uv add advancedresearch

# Or create a new project with advancedresearch
uv init my-research-project
cd my-research-project
uv add advancedresearch

Alternative Installation Methods

BASH
# Using pip
pip install advancedresearch

# Using poetry
poetry add advancedresearch

Development Installation

For development or to access the latest features:

BASH
# Clone the repository
git clone https://github.com/The-Swarm-Corporation/AdvancedResearch.git
cd AdvancedResearch

# Install with uv (recommended)
uv sync

# Or with poetry
poetry install

# Or with pip
pip install -e .

Why uv?

We recommend uv for the best experience with AdvancedResearch:

• ⚡ 10-100x faster than pip for dependency resolution and installation
• 🔒 Reliable: Deterministic builds with automatic virtual environment management
• 🎯 Simple: Single tool for project management, dependency resolution, and Python version management
• 🔄 Compatible: Drop-in replacement for pip with better performance

Environment Setup

Create a .env file in your project root:

BASH
# Claude API Key (Primary LLM)
ANTHROPIC_API_KEY="your_anthropic_api_key_here"

# Exa Search API Key
EXA_API_KEY="your_exa_api_key_here"

# Optional: OpenAI API Key (alternative LLM)
OPENAI_API_KEY="your_openai_api_key_here"

🚀 Quick Start

Complete uv Workflow

Get started with AdvancedResearch using uv for the optimal experience:

BASH
# Install uv (if not already installed)
curl -LsSf https://astral.sh/uv/install.sh | sh

# Create a new project
uv init my-research-project
cd my-research-project

# Add advancedresearch
uv add advancedresearch

# Create your research script
cat > research.py << 'EOF'
from advancedresearch import AdvancedResearch

# Initialize the system
research_system = AdvancedResearch()

# Run research
results = research_system.research(
    "What are the latest developments in quantum computing?",
    export=True,
    export_path="quantum_computing_report.md"
)

print(f"Research completed! Report: {results['research_metadata']['exported_to']}")
EOF

# Run your research
uv run research.py

Example

PYTHON
from advanced_research import AdvancedResearch

# Initialize the system
research_system = AdvancedResearch(max_iterations=1)

# Run research
results = research_system.research(
    "What are the latest developments in quantum computing?",
    export=True,
    export_path="quantum_computing_report.md",
)

print(results)

🤝 Contributing

This implementation is part of the open-source swarms ecosystem. We welcome contributions!

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-research-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-research-feature)
5. Open a Pull Request

Development Setup with uv

BASH
# Clone and setup development environment
git clone https://github.com/The-Swarm-Corporation/AdvancedResearch.git
cd AdvancedResearch

# Install development dependencies with uv (recommended)
uv sync --dev

# Run tests
uv run pytest

# Run linting
uv run ruff check .
uv run black --check .

# Run type checking
uv run mypy advanced_research/

# Format code
uv run black .
uv run ruff check --fix .

📄 License

This project is licensed under the MIT License. See the LICENSE file for details.

📚 Citation

If you use this work in your research, please cite both the original paper and this implementation:

BIBTEX
@misc{anthropic2024researchsystem,
    title={How we built our multi-agent research system},
    author={Anthropic},
    year={2024},
    month={June},
    url={https://www.anthropic.com/engineering/built-multi-agent-research-system}
}

@software{advancedresearch2024,
    title={AdvancedResearch: Enhanced Multi-Agent Research System},
    author={The Swarm Corporation},
    year={2024},
    url={https://github.com/The-Swarm-Corporation/AdvancedResearch},
    note={Implementation based on Anthropic's multi-agent research system paper}
}

@software{swarms_framework,
    title={Swarms: An Open-Source Multi-Agent Framework},
    author={Kye Gomez},
    year={2023},
    url={https://github.com/kyegomez/swarms}
}

🔗 Related Work

• Original Paper - "How we built our multi-agent research system" by Anthropic
• Swarms Framework - The underlying multi-agent AI orchestration framework
• Full Documentation - Comprehensive API reference and advanced usage guide

📞 Support

• Issues: GitHub Issues
• Discussions: GitHub Discussions
• Discord: Join our community

<p align="center"> <strong>Built with <a href="https://github.com/kyegomez/swarms">Swarms</a> framework for production-grade agentic applications </strong> </p>

Source: https://github.com/The-Swarm-Corporation/AdvancedResearch