Multi-Agent Coding System

Stale
GitHub Python Apache-2.0

Description

A multi-agent coding system where orchestrator, explorer, and coder agents collaborate on software tasks with shared context.

Key Features

  • Three specialized agent roles: Orchestrator, Explorer, and Coder collaborating on tasks
  • Orchestrator handles task decomposition, dispatch, and verification without touching code
  • Smart context sharing mechanism reuses discovered knowledge across tasks
  • Ranked #13 on Stanford TerminalBench, outperforming Claude Code
  • Supports different LLM models for orchestrator and subagents
  • Fully async architecture with distributed Docker environment management

Use Cases

💡 Automate complex software engineering tasks end-to-end
💡 Perform systematic codebase exploration and analysis
💡 Execute multi-step code refactoring and bug fix tasks
💡 Run automated testing and verification in terminal environments
💡 Build scalable multi-agent coding workflows

Tags

coding-agent multi-agent orchestration terminal python

Categories

💻 Coding Agent 🧠 Multi-Agent

Quick Start

1. Clone the repo: git clone https://github.com/Danau5tin/multi-agent-coding-system
2. Install dependencies: pip install -r requirements.txt
3. Configure your LLM API key
4. Run the orchestrator: python -m orchestrator --task "your task description"
5. The system automatically dispatches explorer and coder agents
Visit GitHub

Related Projects