aizen
aizen is an open-source macOS workspace manager designed to streamline the use of AI coding agents by integrating git worktree management with parallel agent sessions. It provides a native desktop environment where developers can create, manage, and monitor isolated worktrees—each linked to a different branch—while running multiple AI agent tasks concurrently. This approach ensures that agent activities remain sandboxed, preventing cross-branch interference and enabling experimentation at scale. aizen's architecture includes a graphical user interface for visual oversight, a command-line interface for power users, and a web dashboard for remote access and analytics, making it a versatile hub for AI-driven development workflows. Built with Swift and leveraging Xcode, aizen takes full advantage of macOS-native technologies to deliver a responsive and intuitive experience. The application allows users to spawn and control AI agents across multiple worktrees, track progress in real time, and orchestrate complex multi-step tasks. It supports integration with popular AI coding tools and agents through a flexible plugin system or CLI hooks, enabling it to fit into existing developer toolchains. The worktree model eliminates the need for manual branch switching or stashing, and the session manager keeps state synchronized across concurrent operations. Under the hood, aizen handles the heavy lifting of git operations—creating, updating, and deleting worktrees—while providing a unified interface to interact with AI agents. It includes features such as task queues, agent output diffing, and automatic cleanup of stale worktrees. The web component, built separately, offers team-level visibility, allowing project leads to monitor multiple developers' agent sessions from a browser. Additionally, aizen ships with a CLI tool that can be used in scripts or CI/CD pipelines, further extending its utility. As an open-source project with a growing community, aizen aims to become the go-t
- Category
- AI Agents
- Pricing
- Open Source
Problem solved
Software developers and AI/ML engineers often need to run multiple AI coding agents on different branches or variations of a codebase simultaneously, but manually juggling git branches and isolated environments is error-prone, time-consuming, and lacks unified monitoring. aizen solves this by providing a dedicated macOS workspace that leverages git worktrees to create lightweight, isolated copies of a repository for each agent session. This allows parallel, conflict-free agent execution, centralized progress tracking, and easy cleanup, thereby boosting productivity and reducing context-switching overhead.
Use cases
- Running multiple AI code generation agents on separate feature branches to compare outputs or implement different parts of a system simultaneously.
- Conducting prompt engineering experiments by spawning identical codebases across worktrees and executing different agent instructions in parallel.
- Managing long-running agent tasks, such as refactoring or code migration, while continuing development on other branches without interruption.
- Setting up sandboxed environments for AI agents to explore code changes without affecting the main working tree, with easy rollback via worktree deletion.
- Orchestrating a multi-agent DevOps pipeline where one agent generates boilerplate, another writes tests, and a third reviews code—all under unified session management.
Audience
- Developers
- ML Engineers
Pros
- Native macOS integration provides a smooth, responsive user experience.
- Isolated git worktrees eliminate branch-switching and merge conflicts during parallel agent sessions.
- Unified dashboard and CLI allow flexible monitoring and control of multiple agent tasks.
- Open-source with an active community, encouraging rapid iteration and customization.
- Reduces cognitive load by automating worktree lifecycle and session coordination.
Cons
- macOS-only, excluding Windows and Linux developers.
- Requires familiarity with git worktrees, which may present a learning curve for some users.
- Relatively new tool with evolving documentation and limited third-party integrations.
Official website: https://github.com/vivy-company/aizen
Canonical LambdaBase page: https://www.lambdabase.com/tools/aizen