Vibe Kanban Alternatives: What to Use Now That It Is Community-Maintained

You may not need a Vibe Kanban alternative at all. The app still runs locally, its local workspaces keep working, and the repository is still open source under the Apache 2.0 license. What went away when its company closed was the hosted part: shared issues, comments, projects and organizations in the cloud. So the right replacement depends on the job you hired it for. If it was running coding agents side by side and reviewing their diffs, look at Claude Code Projects or Linear’s coding sessions. If it was planning agent work next to the code, look at Backlog.md. If it was keeping a team and several AI apps on the same page, a shared board such as fenbs fits better. Each is described below, with who should pick it.

What changed at Vibe Kanban

On April 10, 2026, Louis Knight-Webb of Bloop announced↗ (opens in a new tab) that the company was shutting down and that Vibe Kanban would live on as open source and community maintained. The post gave the reason directly: most users were on the free tier, and the team could not find a business model it wanted to pursue. It also set out what happened next. The latest version includes a data export feature; remote services stayed available for 30 days, after which Vibe Kanban moved to a fully local architecture; the hosted kanban issues, comments, projects and organizations were removed; and local workspaces continue to function.

The Vibe Kanban repository↗ (opens in a new tab) now opens with “Vibe Kanban is sunsetting” and a link to that announcement, and it still links a guide to self-hosting the cloud part with Docker Compose. The repository was still receiving commits in September 2026. Check it for the current state before you decide.

Do you need to move?

• You ran it alone, on your own machine, for workspaces and diff review: probably not. That part is local and keeps working. The risk is slower fixes as agents and their CLIs change, so watch the repository.
• Your team shared issues through the hosted cloud: yes, or self-host. The self-hosting guide↗ (opens in a new tab) covers deploying the cloud part on your own Linux server with Docker Compose, a domain and GitHub or Google sign-in. That is a server someone now has to run.
• You are choosing a tool today: pick by job, below, and treat a community-maintained project as one option among several.

If the job was running agents in parallel

Vibe Kanban’s core was the workspace: an agent such as Claude Code in its own git worktree, on its own branch, with a terminal, a dev server and a diff you could comment on. Two hosted products now cover much of that, from the agent vendor and from the tracker.

Claude Code Projects, in beta since September 17, 2026, is Anthropic’s version. You give work to one project conversation; Claude starts a thread for each task, and per the Claude Code documentation on Projects↗ (opens in a new tab) each cloud thread works on its own branch and opens a pull request when the work calls for one. An Overview pane shows which threads finished, which pull requests are ready for review and which thread is waiting on you, and all threads share a project memory. It is on Pro and Max plans, for one user, with code on github.com. It runs Claude only. Details of its memory are in Claude Code Projects shared memory.

Linear’s coding sessions↗ (opens in a new tab) start from the tracker instead. When you delegate an issue to Linear, it starts a secure coding session through Claude Code or Codex in a managed sandbox, drafts a pull request and adds the diff to the issue for you to check before requesting review. Coding sessions are on Linear’s Basic, Business and Enterprise plans and draw on the workspace’s AI credits. This suits a team whose issues already live in Linear. How Linear’s agents compare with Jira’s is in Rovo vs Linear agents.

Neither is a drop-in replacement. Vibe Kanban let you switch between more than ten agent CLIs on your own machine; Claude Code Projects runs Claude in the cloud, and Linear’s sessions run in Linear’s sandbox. If running several different agents locally is the point, staying on Vibe Kanban, or running Claude Code tasks in parallel with worktrees yourself, may serve you better.

If the job was planning agent work next to the code

Some people used Vibe Kanban mainly as a board for agent tasks. Backlog.md does that without a server: every task is a Markdown file in your repository, backlog board draws a kanban board in the terminal, backlog browser serves a local web board, and agents read and update tasks through its CLI or an MCP server. It is MIT-licensed and needs no account. It does not run agents for you; you run them and they work from the tasks. What it is good at, and when it stops fitting, is in Backlog.md alternatives.

If the job was keeping people and AI apps on one page

Vibe Kanban’s hosted cloud was also where a team saw each other’s issues. If that is the part you miss, the question is no longer how to run agents. It is where the shared record lives: what the team agreed to build, the rules agents must follow, what was decided, and where the last session stopped.

fenbs is a web app built for that. People and AI apps read and update the same kanban board over MCP: Claude Code, Codex, Cursor and Copilot in the editor, and Claude or ChatGPT in the browser through a custom connector. Each connection acts as the person who approved it, under that person’s role. Rules go on the Decisions and rules page and come first in every assistant’s AI context; assistants follow them and cannot sign off decisions or pre-approve work. Saying “save progress” records where each task stands and the project’s Where we left off. History records who changed what, a person or which assistant.

What fenbs does not do is the part Vibe Kanban was famous for. It does not start agents, create worktrees, show diffs or open pull requests. It has four fixed lanes, To Do, Next Up, In Progress and Completed, with no review lane and no custom columns. Review is a habit you write down: the assistant comments what changed and how it was checked, and a person moves the task to Completed. The two combine well, as Vibe Kanban and Claude Code describes: the shared board carries outcomes and decisions, the agent tool carries runs.

Side by side

• Vibe Kanban: local app, open source, community-maintained; runs many agent CLIs in worktrees with diff review; hosted cloud removed, self-hosting possible.
• Claude Code Projects: hosted by Anthropic, beta on Pro and Max; Claude threads on their own branches with shared project memory; one user per project.
• Linear coding sessions: inside Linear; delegate an issue and get a draft pull request from Claude Code or Codex; for teams already on Linear.
• Backlog.md: Markdown task files in your repository with a terminal and local web board; no server, no account; does not run agents.
• fenbs: hosted shared board for people and every AI app over MCP, with roles, rules, saved progress and history; does not run agents.

Moving your issues out

If you decide to move, start with Vibe Kanban’s own export, which the shutdown post says the latest version includes. To bring the list into fenbs, use Add many next to + New task on the board: paste one task per line, or a Markdown list, and start a line with bug:, feature: or enhancement: to set its kind. Everything is previewed before anything is created. Bring over the work that is still open and leave the finished history where it is.