Verified project record
chaitanyagiri/munder-difflin
munder-difflin is a local desktop harness that turns existing terminal-agent CLIs into a coordinated office of persistent agents managed through one orchestrator and a visual interface. It runs each configured agent as a real process in its own pseudo-terminal and lets a single supervisor agent assign work and route traffic.
Project overview
Instead of simulating agents, it wraps real terminal-agent CLI processes into a hive supervised by one orchestrator, represents them on an interactive office floor, and retains agent knowledge across sessions in searchable markdown memory.
- Project type
- AI Agent · AI Coding · Workflow
- Use cases
- Coding & Development · Automation
- Deployment
- Refer to project documentation
- License
- MIT
Best for
- Developers who already use one or more terminal-agent CLIs and want them coordinated by a single supervisor agent with a visual interface.
- AI engineers who want persistent agent context retained across sessions in searchable markdown memory.
Key capabilities
- Wraps terminal-agent CLI processes into a coordinated hive and provides one supervisor agent that assigns work, routes traffic, and escalates selected decisions to the user.
- Runs each configured agent CLI or custom command as a real process in its own pseudo-terminal rather than simulating terminal output.
- Displays agents as avatars on an office floor and lets the user open a desk, read its live terminal, and type into the session.
- Lets users constrain agent autonomy, require approval for spend, scope, and destructive operations, and stop agents that loop or run away.
- Provides a file tree, editor tabs, save operations, commit history, diffs, branch comparison, and guarded checkout within the application.
- Plans Telegram and richer chat integrations that send channel messages to the orchestrator queue and route replies back to the channel.
Limitations and risks
- The application requires at least one supported agent CLI on the user's PATH; supported choices include Claude Code, Antigravity, OpenAI Codex, xAI Grok, Kimi Code, Gemini CLI, Qwen, OpenCode, Crush, pi.dev, GitHub Copilot, and Cursor.
- The project describes itself as pre-release software with a large surface area.
- Official builds send anonymous usage events unless the user opts out through settings, the DO_NOT_TRACK environment variable, or a source build without the telemetry key; check this setting before adoption if usage events are a concern.
Getting started
- Download and install the latest desktop release, open the application, complete the onboarding wizard, and select Add agent to start the first session. Prebuilt desktop releases are provided for macOS, Windows, and Linux; the standard installation route does not require Node, a toolchain, or a repository checkout, although at least one agent CLI must be available on your PATH.
Evidence and sources
- README: Munder Difflin is a desktop app that wraps **real terminal-agent CLIs** as fully-capable agents, wires them into a **hive mind**, and puts **your clone** in charge — Michael, the…
- README: **Free, open source and performant** — a multi-agent harness that works with the subscriptions you already pay for, on their hourly limits. It turns the terminal coding CLI you al…
- README: 2. **Agents collaborate through the hive** — a local git repo of plain files. They write to their own `outbox/`; the harness's router delivers into recipients' `inbox/`. No agent…
- README: Michael is your clone and the only agent you brief. He assigns the work, routes the traffic, and escalates the few things that actually need you.
- README: Every agent keeps markdown memory that is mined into a shared, searchable palace. Close the app, come back tomorrow, and they still know what they learned.
AI Search
Find projects, verify facts, compare options, or turn a complex need into an actionable plan
Try a searchA click only fills the search box; you stay in control
Project Details
0