chaitanyagiri/munder-difflin

A local multi-agent harness that works with your existing Claude Code, Codex subscriptions, allows you to run an office of agents

What it solves

Munder Difflin is a multi-agent harness that allows users to run a coordinated team of AI coding agents on their local machine. It solves the problem of managing multiple disparate terminal-based AI CLIs (like Claude Code, GitHub Copilot, or Grok) by wrapping them into a single, self-coordinating "hive mind" where agents can communicate, share memory, and be managed by a single orchestrator agent.

How it works

The system operates by wrapping real terminal-agent CLIs as processes using node-pty and rendering them via xterm.js.

  • The GOD Agent: A central orchestrator (named Michael) acts as the primary interface for the user, assigning tasks, routing messages, and escalating only critical decisions to the human user.
  • The Hive: A local git repository of plain files serves as the coordination layer. Agents communicate via atomic-file mailboxes (inboxes/outboxes) and a shared blackboard.
  • Memory Layer: A markdown-first memory system with a semantic recall index allows agents to remember information across different sessions.
  • Visual Interface: A Pixi.js-powered 2D office floor visualizes agents as avatars who move to stations and send "envelopes" (messages) to one another.

Who it’s for

Developers and power users who already use various AI coding CLIs and want to automate complex workflows by having multiple agents work in parallel, coordinated by a single supervisor agent.

Highlights

  • Broad CLI Support: Wraps Claude Code, Gemini, OpenAI Codex, xAI Grok, Kimi Code, Qwen, OpenCode, GitHub Copilot, and Cursor.
  • Local LLM Integration: Supports bring-your-own-keys and local models via Ollama, LM Studio, or vLLM.
  • Autonomous Coordination: The GOD agent handles routine requests and routing, reducing the need for the user to manage every individual agent.
  • Safety Controls: Includes "human gates" for destructive operations and a circuit breaker to stop agents that loop or exceed budgets.
  • Integrated Tooling: Features a built-in Monaco IDE, Kanban task tracking, and a semantic memory palace for cross-agent recall.

Related

  • Dispatch
    Munder Difflin: Open-Source Multi-Agent Harness for Local AI Clones

    Munder Difflin is an open-source multi-agent harness that wraps existing CLI agent subscriptions to create local, autonomous AI clones of team members that can coordinate work via end-to-end encrypted messaging.

  • Project
    YoanWai/agent-manager

    A terminal-based manager for AI coding agents that consolidates multiple CLI agents into a single interface with live status tracking and integrated diff reviews.

  • Project
    aannoo/hcom

    A CLI tool that enables different AI coding agents to message, observe, and spawn each other across terminals to coordinate complex development pipelines.

  • Project
    SeemSeam/claude_codex_bridge

    CCB (Claude‑Codex‑Bridge) is a cross‑platform terminal UI that lets you run many LLM‑backed CLI agents (Codex, Claude, Gemini, etc.) together in visible panes, define collaboration graphs, share a project‑wide memory file, and even control the workspace from an Android app. Install via npm, configure with a built‑in UI, and manage providers, roles, and rich‑media terminals from the command line.

  • Project
    omnigent-ai/omnigent

    Omnigent is an open‑source meta‑harness that unifies many LLM coding agents (Claude Code, Codex, Cursor, etc.) under one CLI and web UI, letting you run, combine, and govern them across devices and cloud sandboxes, with built‑in collaboration and policy features.