agents-deck vs Vibe Kanban

An honest comparison, row by row, following the comparison table on our landing page. As of August 2026. Corrections welcome.

Vibe Kanban is an open-source (Apache-2.0), self-hosted web kanban for coding agents, and the most agent-agnostic tool in the category: Claude Code, Codex, Cursor, Copilot, Gemini, Amp, Factory and more, over ten in total. That is more integrations than agents-deck supports today, and it deserves to be said plainly.

Two honest caveats before the rows. First, Bloop, the company behind it, shut down in early 2026, and the project is now community-maintained. Second, and more important for us: Vibe Kanban runs agents with --dangerously-skip-permissions (the --yolo flag) by default, which reviewers have flagged as a security concern. That default is the single clearest philosophical difference between the two tools.

Live observability dashboard

Vibe Kanban: partial. A kanban board is a good queue: tasks move across columns as agents pick them up. It is less good as an operations view, and it will not tell you at a glance which agent is stuck waiting for a human.

agents-deck: yes. A live dashboard across agents, tools, and projects, with awaiting-input alerts, built to be the second screen you glance at.

Cross-tool (Claude · Codex · Cursor…)

Vibe Kanban: yes. Genuinely agent-agnostic, with the widest tool support in the category. On raw integration count, Vibe Kanban wins this row.

agents-deck: yes. Claude Code, Codex, Cursor, Kimi, across multiple repos, with the same process guardrails applied to every one of them. One deck.

Human approval, not `--yolo`

Vibe Kanban: no. Agents run with --dangerously-skip-permissions by default. That is fast, and it is exactly the trade we refuse to make.

agents-deck: yes. Your agents shouldn’t run with --yolo. agents-deck adds the approvals and guardrails: humans approve, agents that need input raise an alert, and nothing merges itself.

Process guardrails / governance

Vibe Kanban: no. Isolation and review, like most local orchestrators, but no enforcement of how the work happens.

agents-deck: yes. Validators enforce the process itself: skills with evidence, message banners, status-flip read confirmations, per-agent file jurisdiction. Almost no one does this.

Git-native audit · zero database

Vibe Kanban: no. Task state lives in the app, not as a durable record in your repository.

agents-deck: yes. State is YAML in the repo; the audit trail is git log. Zero database. The repo is the storage. Useful the day someone asks you to prove what your agents did, and the EU AI Act era makes that day likelier.

Cross-session memory & handoff

Vibe Kanban: partial. The board persists, but it is a task queue, not a coordination record designed for handoff.

agents-deck: yes. Tickets, messages, changelog, and roadmap live in the repo, so any session, any tool, any teammate can pick up where the last one stopped.

Self-hosted

Both: yes. Vibe Kanban is self-hosted and open source, and so is agents-deck. On this row the two projects simply agree.

Price

Both free and open source. Vibe Kanban is Apache-2.0. The agents-deck core is free and stays free: dashboard, tickets, messages, changelog, guardrails, CLI, MCP.

Choose Vibe Kanban if

You want the widest possible range of agent integrations today.
A kanban queue matches how you think about dispatching work.
You accept autonomous, skip-permissions execution as the default trade for speed.

Choose agents-deck if

You want approvals and human-in-the-loop safety as the default, not an option.
You want process governance, not just task isolation and a final review.
You want your audit trail in git log, not in an application.
Cross-session memory and clean handoffs between sessions or tools matter to you.

Also comparing?

Sources for the Vibe Kanban side: the official site at vibekanban.com, re-checked August 2026, and an independent review of the permissions default, from May 2026. If we got something wrong, tell us and we will fix it.