Desktop IDE for running Claude Code, Codex, OpenCode and other CLI coding agents in parallel, each in its own Git worktree.
kdlbs/kandev
Self-hostable kanban board for running multiple AI coding agents in parallel, with worktree isolation, an integrated review workspace, and PR creation.
- Stars
- 1,071
- Forks
- 178
- Open issues
- 252
- Last commit
- yesterday
What is it?
Kandev is a kanban-style task manager for AI coding agents. Each task can be assigned to an agent from a supported provider such as Claude Code, Codex, GitHub Copilot, or Gemini CLI, and several tasks can run at the same time. Git worktrees keep concurrent agents from conflicting with each other.
Instead of reviewing agent output only in a terminal, Kandev provides a workspace with a file editor, file tree, terminal, browser preview, and git changes panel. Multi-step workflows can use different agents for different stages, such as planning, implementation, and review. Tasks can also be linked to GitHub, GitLab, Jira, Linear, Sentry, or Azure DevOps issues and PRs.
Kandev runs as a native Go binary with the web UI embedded, and can be installed via Homebrew, Scoop, npm, or built from source. Agents can run as local processes, in Docker containers, on SSH servers, in Kubernetes Pods, or in sprites.dev cloud environments. The project states it has no telemetry.
Key features
- 01Kanban boards with drag-and-drop, plus multi-step agentic workflows that mix agents per step
- 02Git worktree isolation so parallel agents don't conflict, with multi-repository tasks supported
- 03Integrated workspace with editor, terminal, file tree, browser preview, and git changes
- 04Runs agents locally, in Docker, Kubernetes, over SSH, or in cloud executors
- 05Task-agent and external MCP support for creating subtasks and managing Kandev from other agents
- 06Automations triggered by schedules or webhooks
Use cases
- Running several coding agents on different issues at once and reviewing their diffs in one place
- Splitting a large task into subtasks or producing multiple PRs from one starting point
- Offloading agent execution to a remote server while controlling tasks from a browser or phone
- Teams that want a shared, self-hosted workflow for AI-assisted development
Quick start
- 1
Install via Homebrew (macOS, Linux)
brew install kdlbs/kandev/kandev - 2
Launch Kandev
kandev
Things to know
- Requires you to install and authenticate at least one external agent CLI before tasks can run.
- Licensed under AGPL-3.0, a copyleft license that requires sharing modified source when offered as a network service.
In collections
Topics
FAQ
- What is kandev?
- Self-hostable kanban board for running multiple AI coding agents in parallel, with worktree isolation, an integrated review workspace, and PR creation. Kandev is a kanban-style task manager for AI coding agents. Each task can be assigned to an agent from a supported provider such as Claude Code, Codex, GitHub Copilot, or Gemini CLI, and several tasks can run at the same time. Git worktrees keep concurrent agents from conflicting with each other.
- Is kandev free to use? What license does it have?
- kandev is open source under the AGPL-3.0 license. Check the license terms before using it in commercial products.
- Is kandev still maintained?
- Actively maintained. The last commit to kandev was on Nov 6, 2026.
- What can I use kandev for?
- Running several coding agents on different issues at once and reviewing their diffs in one place Splitting a large task into subtasks or producing multiple PRs from one starting point Offloading agent execution to a remote server while controlling tasks from a browser or phone Teams that want a shared, self-hosted workflow for AI-assisted development
- How do I install kandev?
- Install via Homebrew (macOS, Linux): brew install kdlbs/kandev/kandev Launch Kandev: kandev
- Where can I get kandev?
- The source code and installation instructions are on GitHub: https://github.com/kdlbs/kandev
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
A desktop and web workspace where installed coding agents work as lasting, named teammates in shared conversations, with task delegation and collaborative memory.
An open-source coding agent that runs as a desktop app, a terminal CLI, or an installable Skill for Codex, OpenCode and DeepSeek Harness.
Local-first desktop workbench that runs multiple coding agents in parallel, each in its own git worktree, with diffs, editor, terminals and PR actions in one window.
A VS Code fork that runs Claude Code, Codex and other agent CLIs side by side, organizing work by project and git worktree feature.
Open-source Electron desktop workspace for running parallel Codex, Claude Code and OpenCode sessions, with visual markdown, diagram and task editing.