SuperAgent

AI Agent self-organizing teams, task allocation, and collaborative CLI tool.

One command → agents form teams, split tasks, coordinate work, deliver results.

Highlights

• 49 CLI commands covering the full agent lifecycle
• 16 open-source projects integrated (577k+ total GitHub stars)
• 178 professional agent roles across 14 departments
• 4 execution engines: DeerFlow, Aider, Browser-use, GPT-Researcher
• Zero-cost offline mode: Ollama / LocalAI / llama.cpp
• Pixel-art dashboard: Real-time agent visualization

Quick Start

pip install superagent-cli
superagent auto "Build a todo app with auth" --command builtin

Core Commands

Integrated Projects (577k+ Stars)

Architecture

User → superagent auto "goal"
  → Orchestrator (category detect + role match)
  → Execution Engine (DeerFlow / Aider / Local)
  → Multi-Agent (each in Git Worktree + tmux)
  → File Output + Dashboard Visualization

Installation

pip install superagent-cli          # Core
pip install superagent-cli[all]      # Everything

Install from Source (development)

git clone https://github.com/superagent-cli/superagent.git
cd superagent
pip install -e .           # Core
pip install -e ".[dashboard]"  # + Dashboard

Optional: Local Model (zero cost)

brew install ollama && ollama serve &
ollama pull llama3.2
superagent local register
superagent auto "Build X" --command local

Configuration

# LLM Provider (pick one)
superagent llm add --name deepseek --api-key $DEEPSEEK_API_KEY --base-url https://api.deepseek.com/v1 --model deepseek-chat
superagent llm default deepseek

# Or use local models (free)
superagent local setup --model llama3.2

Documentation

• Usage Guide — Complete 10-section tutorial
• Evaluation Report — 5W2H1R + SWOT analysis

License

MIT