Vibe Conductor

关于

Vibe Conductor acts as a central Model Context Protocol (MCP) server, streamlining collaboration among multiple AI agents, particularly Claude Code agents, on complex projects. It manages a shared `tasks.md` file, which serves as the single source of truth for task status and details. Agents interact with this file indirectly through robust MCP tool calls, avoiding direct file I/O, while human users can directly edit `tasks.md`. The server intelligently merges external human edits, ensuring data integrity and preserving agent-owned task states. This setup simplifies the agent's workflow into a "todo → in-progress → done" lifecycle, abstracting the complex underlying state transitions for efficient multi-agent and human-AI collaboration.

主要功能

• Shared `tasks.md` file with atomic operations and safe human edit merging
• Comprehensive MCP toolset for managing sessions, tasks, and notes
• 0 GitHub stars
• Multi-agent task coordination via a central MCP server
• Agent-simplified workflow (todo/in-progress/done) over complex file states
• Ensures data consistency with async mutex for mutations and atomic file writes

使用案例

• Managing software development projects where multiple AI agents collaborate on coding, testing, and committing tasks
• Streamlining workflows for human-AI teams by providing a shared, safely editable task list
• Facilitating complex multi-stage projects where tasks transition through implementation, review, and testing phases