This project provides a multi-agent orchestration system for VS Code Copilot, enabling complex software development workflows through intelligent agent delegation and parallel execution. It addresses the limitations of single-agent approaches by breaking down the development lifecycle (planning, implementation, review, commit) into specialized roles, significantly improving context efficiency and developer productivity. The system is designed for developers looking to automate and streamline their coding processes.

How It Works

Copilot Atlas employs a conductor-delegate architecture. A central orchestrator, Atlas, coordinates specialized subagents like Prometheus (planner), Oracle (researcher), Sisyphus (implementer), and Code-Review (reviewer). This modular design drastically improves context conservation; instead of a single agent processing the entire codebase, specialized agents focus only on their assigned tasks and relevant files, freeing up substantial token capacity for deeper analysis and reasoning. Agents communicate and delegate tasks using VS Code's custom agent handoff features.

Quick Start & Requirements

• Installation: Clone the repository and copy agent files to the VS Code user prompts directory (e.g., ~/.config/Code/User/prompts/ on Linux).
• Prerequisites: VS Code Insiders (recommended for latest features), GitHub Copilot subscription with multi-agent support.
• VS Code Settings: Enable chat.customAgentInSubagent.enabled and set github.copilot.chat.responsesApiReasoningEffort to "high".
• Documentation: Agent-specific details are available in .agent.md files within the repository.

Highlighted Details

• Context Conservation: Achieves 10-15% context usage compared to 80-90% for monolithic agents, enabling more efficient token utilization.
• Parallel Agent Execution: Supports launching multiple subagents simultaneously for independent tasks, up to 10 agents per phase.
• Test-Driven Development: Enforces a strict red-green-refactor cycle, with tests written before minimal code implementation.
• Proper Agent Handoffs: Utilizes VS Code's custom agent handoff configuration for seamless workflow transitions.
• Structured Planning: Generates detailed, incremental plans with risk assessments and user approval gates.

Maintenance & Community

No specific details regarding maintainers, community channels (like Discord/Slack), or roadmap are provided in the README.

Licensing & Compatibility

The project is released under the MIT License, which permits commercial use and modification with minimal restrictions, requiring only the inclusion of the copyright notice and permission notice.

Limitations & Caveats

The system is best supported on VS Code Insiders, and its functionality is dependent on a GitHub Copilot subscription. The effectiveness of the agents relies on the capabilities and limitations of the underlying AI models (Claude Sonnet, GPT-5.2, Gemini).