Run AI coding agents in parallel
Spin up Claude Code, Cursor, Copilot, Gemini CLI and more at the same time. Each task gets an isolated git worktree with project repo defaults, built-in code review, and MCP + IDE integration.
Install Node.js 18+ then run:
$ npx vibe-kanban
Latest release v0.1.41 (Apr 3, 2026) brings faster big-PR diffs, hide-blocked filters, clickable alerts, and more reliable conversation history.
What’s new
Highlights from v0.1.41 (April 3, 2026).
Faster diffs for big PRs
Virtualized, self-correcting diff streams stay smooth even on 300+ file reviews.
Hide blocked cards + alerts
Filter the kanban board to focus on unblocked work and jump back from desktop notifications.
More reliable conversations
Execution sessions now clean up correctly, preventing chat history leakage across tasks.
Production diff/tab fixes
Portable worker bundles and patched Docker builds keep the diff tab and previews stable.
How it works
From one command to parallel agent workflows in four steps.
Install & run
Run npx vibe-kanban and open the web UI. No account required.
Create tasks
Add tasks and pick an agent (Claude Code, Cursor, Copilot, etc.). Each task gets its own workspace.
Agents run in parallel
Agents work in isolated git worktrees so they never conflict. Monitor progress from one place.
Review & merge
Use built-in diffs to review, edit, and approve changes—then merge when ready.
"Vibe kanban is the biggest increase I've had in productivity since cursor."
Choose your coding agent
Works with Claude Code, Cursor, GitHub Copilot, Gemini CLI, and more. Switch or run multiple agents in parallel.
Built for parallel AI development
Run local coding agents in parallel, give each its own workspace, and review changes like a pull request.
Run agents in parallel
Focus on planning and reviewing while agents work in the background. No more staring at logs.
Learn more →Isolated workspaces
Each task runs in its own git worktree. Agents never step on each other; merge when tasks complete.
Learn more →Code review built in
Inspect, edit, and approve agent changes with built-in diffs—like reviewing a PR.
Learn more →Use cases
From refactors to bug fixes to new features—run multiple AI coding agents at once.
Feature development
Split a feature into tasks and run them in parallel. UI, API, and tests can progress at the same time.
Read more →Refactoring at scale
Let agents tackle different modules or files. Review and merge incrementally.
Read more →Bug fixes & maintenance
Queue multiple bugs or small tasks and let agents work through them in parallel.
Read more →Topic guides
Deep dives for search and onboarding: Claude Code, Cursor CLI, worktrees, MCP, and productivity patterns.
Claude Code in parallel
Run several Claude Code tasks with isolated worktrees and one review surface.
Claude Code workflow →Cursor CLI
Use Cursor as an agent alongside others without branch collisions.
Cursor + Vibe Kanban →Git worktrees
Why worktrees matter when multiple AI agents edit the same repo.
Worktrees explained →Local-first coding
Keep source on your machine; orchestrate agents and keys yourself.
Privacy-friendly setup →Engineering throughput
Productivity patterns when execution parallelizes and review becomes the focus.
Productivity guide →GitHub Copilot in parallel
Multiple Copilot-backed tasks with worktree isolation and one merge queue.
Copilot workflow →Gemini & Codex
Run Google Gemini CLI or OpenAI Codex on separate cards alongside other agents.
Gemini CLI → · Codex tasks →Install & VS Code
From npx vibe-kanban to IDE-native monitoring and review.
Monorepos & teams
Split refactors per package; align leads with a shared AI playbook.
Monorepo guide → · Team playbook →Orchestration & review
Control plane for many agents plus a merge-ready checklist.
Orchestration → · Review checklist →Fits your workflow
MCP, IDE extensions, and GitHub. Thousands of developers use Vibe Kanban daily.
MCP & Claude Desktop
Connect MCP clients like Claude Desktop to create and manage tasks from your AI assistant.
MCP integration →IDE integration
Monitor agents, do code review, and see running processes from your IDE.
IDE guide →"This project is underrated. I've found it to be useful and fun."
Frequently asked questions
Is Vibe Kanban free?
Yes. Vibe Kanban is free and open source. You only pay for the AI services (e.g. Claude, OpenAI) you use with your agents.
Which coding agents are supported?
Claude Code, Cursor CLI, GitHub Copilot, OpenAI Codex, Gemini CLI, Amp, Opencode, Factory Droid, and more. See supported agents.
Is my code sent to your servers?
No. Vibe Kanban runs locally. Your code stays on your machine; agents run in your environment.
Why we built it
Coding agents are reliable enough to be the default
We've built and tested coding agents for years. Reliability has crossed a threshold where we prefer starting most tasks with them.
Agents will keep getting better
We expect today's failure modes to shrink over time. We're building tools for the world where engineers spend more time on design and review.
The best agent will keep changing
You shouldn't have to change your workflow to try different coding agents. Vibe Kanban lets you switch or combine them.
Ready to get started?
Run AI coding agents in parallel with one command.
Install Node.js 18+ then run:
$ npx vibe-kanban