Isolated project sessions
Runs agents in a Linux micro-VM on a copy-on-write clone, while keeping the real project folder out of the session.
Mudroom
Mudroom runs coding agents in an isolated Linux VM or Docker/Podman environment on a clone of a project, then shows their changes as a diff for selective review and application. It supports several coding agents and command-line workflows across macOS, Linux, and Windows.
Run coding agents in an isolated project copy
Mudroom is a developer tool for running coding agents against an isolated copy of a project. It places an agent in a Linux micro-VM or a Docker/Podman-based environment rather than mounting the user's real folder, which is intended to make permission-disabled agent workflows easier to review before they affect the source project.
After an agent finishes, Mudroom presents the resulting changes as a diff. Users can apply the complete change set, selected files, individual hunks, or nothing, and can undo an application later. The project is an early prototype, so its documentation warns that rough edges and breaking changes are possible.
Key features
Controlled network access
Restricts outbound access to hosts the user allows, with a network-check command for Docker and Podman environments.
Multiple agent options
Supports Claude Code, Codex, Gemini CLI, opencode, Aider, and arbitrary commands, with agent sign-in and API-key setup.
Selective review and application
Presents completed work as a diff and lets users apply all changes, selected files, selected hunks, or no changes.
Cross-platform operation
Provides a macOS app and CLI on Apple-silicon Macs, plus a CLI for Linux, Windows, and Intel Macs using Docker or Podman.
Configurable sandbox backends
The CLI can select Apple, Docker, or Podman backends explicitly or automatically, and sessions remember their backend.
Use cases
Safer autonomous coding sessions
Run an agent with permission prompts disabled while keeping it away from the working folder, then inspect the proposed patch before changing the project.
Network-constrained development
Give an agent a bounded set of approved network destinations when tasks require package downloads, APIs, or other remote resources.
Incremental patch review
Use the diff view to accept a complete implementation, select only particular files or hunks, or reject the result without modifying the project.
Cross-platform agent execution
Run the same general workflow from the macOS app or CLI, or use Docker and Podman on Linux, Windows, and Intel Macs where the app is unavailable.
Pros and Cons
Pros
- Keeps the real project folder out of the agent session.
- Supports selective, diff-based review before changes are applied.
- Offers configurable network allowlisting and multiple container or VM backends.
- Works with several named coding agents as well as arbitrary commands.
- Supports macOS, Linux, Windows, and Intel Macs with platform-specific runtime requirements.
Cons
- The project is an early prototype and may include rough edges or breaking changes.
- The macOS application is ad-hoc signed rather than notarized, and Windows binaries are not code-signed.
- The graphical app is limited to Apple-silicon Macs; other supported platforms use the CLI.
FAQ
What platforms and runtimes does Mudroom support?
Mudroom runs coding agents in a Linux micro-VM on Apple-silicon Macs, and through Docker or Podman on Linux, Windows, and Intel Macs. The real project folder is not mounted into the session.
How do I install and set up Mudroom?
On an Apple-silicon Mac running macOS 26 or later, use the install script or Homebrew cask, complete setup, build the agent image, sign in to an agent, and create a session. Linux, Windows, and Intel Mac users use the CLI with Docker or Podman.
Which coding agents can Mudroom run?
The README lists Claude Code, Codex, Gemini CLI, opencode, and Aider, and also says Mudroom can run any command. Agent sign-in and API-key configuration are part of the setup workflow.
How are agent changes reviewed and applied?
When a session ends, Mudroom shows the changes as a diff. You can apply all changes, selected files, selected hunks, or none, and undo an apply later.
What limitations should users know about?
Mudroom is identified by its maintainers as an early prototype, so users should expect rough edges and breaking changes. The macOS application is ad-hoc signed and not notarized, and Windows binaries are not code-signed.
Quick Facts
- Category
- Developer tool
- Primary workflow
- Run an agent, review its diff, then selectively apply changes
- Supported agents
- Claude Code, Codex, Gemini CLI, opencode, Aider, or any command
- Platforms
- macOS 26+ Apple silicon, Intel Mac, Linux, and Windows 10/11
- Sandbox backends
- Apple container, Docker, or Podman
- Project status
- Early prototype
Alternativas a Mudroom
ByteAsk
ByteAsk is a terminal-first AI coding agent for C and C++ that edits repositories and verifies changes with the real compiler, debugger, sanitizers, and tests before showing a diff. It offers a free tier plus paid plans, with editor connectors and zero-retention handling described in the source.
GitBot
GitBot is a local browser-based workspace for creating reusable coding bots powered by Claude Code, Codex, or OpenCode. It helps developers run those agents across repository folders with persistent threads and configurable permissions.
Ghost
Ghost es un asistente de IA para terminal para chatear, generar código y ejecutar tareas en la línea de comandos. Incluye modelos gratuitos, es compatible con Linux, macOS y Windows, y es de código abierto.
Vi3ecode
Vi3ecode is a maintained development environment for working with AI coding agents. It keeps project context, terminal, Git, memory, flows, review, and communication connected around the agent you already use.
CreateOS Sandbox
CreateOS Sandbox is an isolated compute environment for running code and agent workloads inside Firecracker micro-VMs. It is designed for workflows that need machine-level isolation, private networking between sandboxes, and programmatic control through SDK, CLI, or MCP.
hob
hob is an independent workspace for coding agents that keeps agent sessions, terminals, history, and follow-up work organized around the tools and providers you already use. It is aimed at developers who want local control over routing, history, and workspace structure rather than a bundled model stack.