HarnessMD/munder-difflin

an open-source alternative to the dots, bots and muses of the world, run an office of claude code/codex like agents on your laptop, sandboxes or anywhere,...

View on GitHub ↗Jump to charts ↓Open shareable report →

HarnessMD/munder-difflin has 8.5k+ stars and 1k+ forks on GitHub, is written in TypeScript, and was last pushed on Oct 18, 2026. Data as of . Signed-in members get hourly updates — create a free account.

Summary Information

Updated 33 minutes ago
Added to GitGenius on October 2nd, 2026
Created on May 31st, 2026
Open Issues & Pull Requests: 251 (+0)
GitHub issues: Enabled
Number of forks: 1,227
Total Stargazers: 8,866 (+2)
Total Subscribers: 35 (+0)

Repository Insights (GitGenius)

Median issue/PR response: 21.0 hours
Mean response time: 6.0 days
90th percentile: 16.3 days
Tracked items: 131

How this project is maintained

Roughly one issue in three opened in the past year never receives a reply. 76% of open issues come from outside the core team, so the backlog reflects real-world use rather than internal planning. Almost all tracked open issues have seen activity in the last three months. Only 46% of issues opened in the past year have been closed. Three people close 90% of everything that gets resolved.

Charts & Analytics

Fetching additional details & charts...

Issue Activity (beta)

Open issues: 120
New in 7 days: 6
Closed in 7 days: 0
Avg open age: 27 days
Stale 30+ days: 72
Stale 90+ days: 1

Recent activity

Opened in 7 days: 6
Closed in 7 days: 0
Comments in 7 days: 2
Events in 7 days: 4

Top labels

  • bug (72)
  • enhancement (47)
  • good first issue (11)
  • ui (9)
  • os:windows (7)
  • hive-protocol (6)
  • os:macos (6)
  • providers (5)

Most active issues this week

Sign in to see which issues are moving.
Sign in

Detailed Description

Munder Difflin is a multi-agent harness that orchestrates multiple AI coding agents on your local machine.

The tool solves the problem of running multiple AI coding assistants in parallel while respecting their hourly rate limits. It wraps existing CLI tools and subscriptions you already pay for—including Claude Code, Gemini CLI, OpenAI Codex, xAI Grok, Kimi Code, and others—and coordinates them as a team of agents working together. A primary agent called Michael acts as the orchestrator, routing tasks between agents, managing their memory, and visualizing their work as avatars on a shared office floor. The system runs locally and works with bring-your-own API keys or local language models.

The tool suits developers who want to maximize the value of multiple AI coding subscriptions by running them concurrently on a single machine. It works best for projects where you can decompose work into parallel tasks that different agents can handle simultaneously. The local-first approach means no data leaves your machine, and the free, open-source nature removes licensing barriers to adoption. The desktop application provides a visual interface showing agent activity in real time rather than requiring terminal-only interaction.

The project maintains active development with regular updates and a changelog tracking changes. The maintainers welcome community contributions and provide documentation including guides and postmortems on the blog. The tool runs on macOS, Windows, and Linux, and is built with Electron, React, TypeScript, and specialized libraries for terminal emulation and graphics rendering.