Kandev CLI
A workspace for parallel Claude Code and Codex tasks, with separate Git worktrees, terminals and change review
- 1,067 ★ on the repository
- Last commit
- AGPL-3.0
Broad access
Why this level
- Executors edit code and run commands in their configured environments
- Integrations can open pull requests and publish conversation snapshots
- Profiles and worktrees can receive secrets and environment files
Install
- Works in
Manual install
brew install kdlbs/kandev/kandev
kandevThe README’s first quick start installs native Kandev and starts its server and web workspace on macOS/Linux. Connect agent CLIs separately. We did not run these commands.
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- Executors edit code and run commands in their configured environments
- Integrations can open pull requests and publish conversation snapshots
- Profiles and worktrees can receive secrets and environment files
This is third-party code. Review the repository files before installing.
What it does
Kandev starts a local server with a web workspace for several coding agents. You create tasks and define workflow stages, then choose the agent and launch profile for each task. A task can span multiple repositories; Kandev prepares a Git worktree for each. Conversations appear beside terminals and file changes, so you can leave feedback or switch to the agent’s familiar interface. Kandev works with Claude Code, Codex and other CLIs. Through MCP, agents create subtasks, read plans and exchange messages. An external agent running outside Kandev can connect to the same board.
Who it is for. For developers and project leads who run several agent tasks concurrently and review changes before merging.
Good fit when
- Several tasks run concurrently and each needs its own branch and discussion history
- A change spans multiple repositories or needs several pull requests
- A team wants defined handoffs from planning to implementation and review
- You need to review agent output alongside a terminal and application preview
Not a fit when
- You only need a short session in your existing terminal without a shared board
- You need Git alone to confine an agent’s permissions to its project folder: worktrees do not restrict process permissions
- You need an autonomous team in Office mode: it is still in development and is not a supported feature
Example request
Create a Kandev task for updating authentication across two repositories, prepare separate branches and show changes before opening pull requests.Limitations
Chosen agents need their own accounts and model access; Kandev supplies no subscriptions or keys. npm launch requires npm 7 or later, while Homebrew and Scoop bundles run without Node.js. Executor profiles set launch commands and unattended permissions; inspect the profile before starting a task. An .env copied into a worktree is readable by the agent, and code or conversations it reads may reach the cloud-model provider.
How to disable. Stop Kandev and its running tasks, then uninstall it through your package manager, for example `npm uninstall -g kandev`. Inspect repositories, worktrees and backups separately: removing the program does not undo commits or published pull requests.
FAQ
How do I install Kandev?
For a one-off launch, run `npx kandev@latest`. For a persistent installation, use `npm install -g kandev@latest`, then start the application with `kandev`.
Can I connect an agent without a dedicated adapter?
Yes. Run its CLI through PTY passthrough and use its familiar terminal interface. This route does not require a dedicated adapter.
How can another team reuse a development workflow?
Export the workflow as YAML and import it into another Kandev workspace. The file contains stages and settings; agent and repository access must be configured separately.
Can an external agent manage the board?
Yes. In Kandev settings, choose the MCP configuration snippet for your client, such as Claude Code, Cursor or Codex. The connection uses Streamable HTTP or SSE and exposes tools for reading tasks, creating subtasks and completing stages.
Can I share a task conversation for review?
Yes. Kandev previews the conversation snapshot before publishing it as a link-accessible GitHub Gist. Review the content first: anyone with the link can see it.
What license does Kandev use?
AGPL-3.0-only. Review its terms before modifying or hosting the service.
Discussion
No questions or reviews yet.
Related
CLI toolsA local control panel for coding agents: it runs Claude Code and Codex, keeps a task queue and merges verified work into main.
A CLI opens the Cezar dashboard for coding-agent tasks, with separate sessions, Git worktrees and subtask reports
DevChain runs a team of AI coding agents on your own machine with a shared board, chat, code review and live terminals.
A bridge between local CLI agents: Codex, Cursor or Claude Code delegate work to Antigravity, Grok, Kimi and seven more.
A memory and coordination layer for coding agents: specs and issues live in git while agents work in parallel
A terminal kanban board that drives several AI agents in parallel with worktree isolation and independent review.
README in short
Kandev creates a data snapshot before an upgrade. System settings let you download a snapshot or restore data from it. Logs and disk-usage information are available there too.