AgentIndex icon
AgentIndex
ToolsMCP ServersHOTCategoriesTrendingNewBest ListsCompareGuides
Submit Tool
ToolsMCP ServersHOTCategoriesTrendingNewBest ListsCompareGuides
Home/
Multi-Agent/
munder-difflin
munder-difflin logo

munder-difflin

Active·★ 9.6k·MIT·Updated 2026-11-17

Munder Difflin is an open-source multi-agent desktop harness (MIT licensed) built with Electron, React, and Pixi.js that reimagines AI agent coordination as a living 2D virtual office floor. Rather than interacting with fragmented command-line tools in isolation, Munder Difflin wraps real terminal-agent CLIs—including Claude Code, Codex, Gemini CLI, Grok, Kimi, and OpenCode—into persistent processes managed through pseudo-terminals (`node-pty`) and rendered via xterm.js.

Orchestration is centered around an autonomous supervisor agent ('Michael' / the GOD agent). Developers brief Michael through a unified command center; Michael evaluates task complexity, allocates tasks to specialized agent desks on the visual floor, and routes inter-agent communications through an asynchronous Git-backed mailbox system (`inbox/` and `outbox/`). A single-committer architecture prevents Git index locking while preserving a permanent, auditable log of code decisions. In addition, the application integrates local on-device Whisper transcription for dictating tasks and capturing meeting notes directly into agent queues.

Binaries are provided for macOS, Windows, and Linux. While providing an engaging visual representation of multi-agent workloads, prospective teams should evaluate operational realities: running multiple concurrent terminal agents rapidly burns through hourly subscription rate limits across commercial AI providers. Furthermore, the combined footprint of Electron, Pixi.js rendering, and background LLM polling creates significant local CPU and RAM utilization.

This write-up is based on the project's README and public repository data. We have not hand-tested the tool. Data as of 2026-11-12.

License
MIT
Latest release
v0.5.5 · 2026-10-01
Last commit
2026-11-11
Contributors
60
Forks
1,336
Open issues and PRs
289
First published
2026-05-31

Source: GitHub, npm and PyPI public APIs.

#agent-framework#agent-harness#agent-orchestration#agents#ai-agents#autonomous-agents#claude-code#codex
↗ Visit site★ GitHub
01

Features

01A visual office and terminals for separately running agent CLIs
02Shared task board with per-agent memory and mailboxes
03Human gates, budgets and optional git worktree isolation
02

Why choose it

+Wraps real existing CLI subscriptions (Claude Code, Codex, Gemini CLI) into unified visual avatars.
+Centralized orchestrator agent autonomously dispatches sub-tasks, routing inter-agent messages via Git mailboxes.
+Built-in offline Whisper model provides local dictation and meeting transcription with zero external API fees.
+Permanent markdown memory palace preserves context and decisions across application restarts.
03

Trade-offs

!Running multiple autonomous agents in parallel rapidly consumes provider hourly rate limits and quotas.
!Desktop Electron and Pixi.js graphics stack carries noticeable baseline memory and CPU overhead.
!Inter-agent communication relies on file-based polling loops, introducing small execution latencies.
!Advanced features like dual-track meeting transcription (Stapler) are reserved for paid Pro licenses.
04

Maintenance check

ActiveChecked: 2026-11-12
Last commit
2026-11-10
Commits, last 90 days
100+
Releases, last 90 days
13
Latest release
v0.5.5 (2026-10-01)
Open issues
289
Recent issues with at least one comment
3/10
Commits per week, last 12 weeks
Commits per week, last 12 weeks: 103, 41, 32, 35, 31, 57, 29, 21, 12, 7, 10, 5

Oldest week on the left; the latest four weeks are orange. Source: GitHub commit activity.

Status rule: archived on GitHub = Archived; last commit within 90 days = Active; 91 to 365 days = Slowing; older = Stale. All figures come from the GitHub API on the date shown.

Most discussed open issues

  • #103 Add other TV series based layout and sprites as themes for agent visualisation4 comments · 2026-07-02
  • #642 Bug: Agent resume crashes: "Permission overrides are not supported when resuming a remote task"3 comments · 2026-09-29
  • #716 Opt-in guardrail: rules agents keep, with a backstop and approve-on-a-card2 comments · 2026-10-08
  • #695 Slack: injected prompt gets "stuck in the input box" — PTY auto-submit fails after multi-line bracketed paste2 comments · 2026-10-04
  • #687 Request: per-seat Unix-socket allowance in the generated seat sandbox settings2 comments · 2026-10-04

Issue titles are quoted from GitHub. Open the link to check whether a problem still applies to the version you use.

05

Compatibility

macOS
Packaged Electron app; agent CLI needed
Verified via docs
Windows
Packaged desktop app
Verified via docs
Linux
Packaged desktop app
Verified via docs
Agent CLIs
Wraps installed Claude Code, Codex and other CLIs
Verified via docs
06

Use cases

↳Coordinating several coding agents around a shared project
↳Inspecting task progress, terminals and agent communication
↳Retaining role-specific context between agent sessions
07

Alternatives

⚡Top VS:
VScezar★1.1k→VSoh-my-agent★1.5k→VSGenAI_Agents★25.0k→VSagentes-chat★1.7k→
cezar logo
cezar★ 1.1k
Open-source orchestrator and ADE for running Claude Code, Codex, OpenCode, Pi and other AI coding agents in parallel — locally or 24/7 on your own server.
VSCompare→
SoduWork logo
SoduWork★ 1.7k
The AI-native engine for teams and their agents. Multi-harness, multi-agent, self-hosted.
oh-my-agent logo
oh-my-agent★ 1.5k
Agent skills and specialist teams for Claude Code, Codex, Cursor and other coding agents. Load one skill or install domain presets, project hooks and parallel workflows with test gates.
VSCompare→
GenAI_Agents logo
GenAI_Agents★ 25.0k
50+ tutorials and implementations for Generative AI Agent techniques, from basic conversational bots to complex multi-agent systems.
VSCompare→
agentes-chat logo
agentes-chat★ 1.7k
Open-source, self-hosted Slack alternative where your team and AI agents work together in channels and DMs. Agents run on Claude, ChatGPT/Codex, Gemini or Grok, with MCP tools, their own computer and browser, routines, memory and approvals. Web, desktop and phone apps. Docker Compose + Postgres. AGPL-3.0.
VSCompare→
SWE-AF logo
SWE-AF★ 1.1k
Autonomous software engineering fleet of AI agents for production-grade PRs on AgentField: plan, code, test, and ship.
VSCompare→
orca logo
orca★ 127.8k
Orca lets you run a fleet of coding agents (e.g. Claude Code, Codex) with your own subscription. Available on desktop, mobile and remote runtime.
paperclip logo
paperclip★ 114.6k
The open-source app everyone uses to manage agents at work
VSCompare→
See all alternatives →

Comments

Log in to leave a comment

No comments yet. Be the first!

On this page
01Features02Why choose it03Trade-offs04Maintenance check05Compatibility06Use cases07Alternatives
HarnessMD /
munder-difflin
↗

Multi-agent office harness turning local coding CLIs into a coordinated 2D workspace with persistent memory

9.6k1.3kMIT
Last commit1d ago
StatusActive
CategoryMulti-Agent
Contributors60
Latest releasev0.5.5
Links
Documentation↗Discussion↗Issues↗Releases↗
Cloud Infrastructure
Ad

Deploy on DigitalOcean

Host your MCP servers, Docker containers, and AI agents with $200 free credit over 60 days.

Claim $200 Free Credit
© 2026 AgentIndex.app|Built by a 10-year iOS Developer.
Best by use caseAboutContactAdvertisePrivacy PolicyTerms of UseQYSGitHubBuy me a coffee ☕

Browse by Category

MCP ServersCode AssistantWorkflow AutomationRAG / Knowledge BaseMulti-AgentBrowser AutomationLLM InfraDev ToolingObservability

Not affiliated with Anthropic, OpenAI or Microsoft.