Kandev

AI kanban and development environment: define tasks on kanban or pipeline boards, dispatch them to coding agents in parallel, then review the changes in an IDE-like workspace and open PRs.

• Maker: kdlbs (GitHub organisation); site kandev.ai. No company note exists.
• Licence / cost: AGPL-3.0, open source, self-hostable, “no telemetry” (README claim).
• Platforms: Go binary for macOS, Linux and Windows (Homebrew, Scoop, npx, release archives) with a web UI; a desktop build is also released.
• Supported agents (README): Claude Code, Codex, GitHub Copilot, Gemini CLI, Amp, Auggie, OpenCode, Cursor, Devin, Qwen, Factory Droid, Kilocode, Pi, Kimi, Kiro, Qoder, Trae, Grok, Hermes, Antigravity and others.
• Features: drag-and-drop boards, multi-step workflows mixing agents per step, git worktree isolation, multi-repo tasks, runtimes (local, Docker, SSH, cloud executors), review workspace (editor with LSP, terminal, browser preview, diffs), automations on schedule/webhook, sub-tasks, task-agent and external MCP, integrations with GitHub, GitLab, Jira, Linear, Sentry, Azure DevOps, mobile access via Tailscale/Cloudflare Tunnel.
• Maturity: repo created 2026-01-09, v0.97.0 on 2026-10-04, roughly 900 GitHub stars at check time (point-in-time). An “Office mode” for persistent agent teams is in progress.
• Versus similar tools: in the author’s view, the most board/pipeline-oriented of this group, and server-style rather than a plain desktop wrapper; compare Ghostex (has a Kanban plugin) and Synara.

Open items

• No independent third-party reviews were checked; the “no telemetry” and feature list are README claims.

Sources (fetched 2026-10-05)

• https://github.com/kdlbs/kandev (README, releases, metadata via GitHub API)