Kangentic: a board for running several coding agents
Kangentic
A desktop board turns a card into an agent session in its own git worktree, flags the agent waiting on you, offers an MCP server
- 142 ★ on the repository
- Last commit
- AGPL-3.0
Extended access
Why this level
- The app launches agents with your credentials and permissions, and automations can call scripts and webhooks
- Anonymous telemetry is on by default in release builds and is switched off with an environment variable
Install
- Works in
Manual install
npx kangenticDownloads the release for your system and opens the app. Requires Node.js 20 and Git 2.26.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The app launches agents with your credentials and permissions, and automations can call scripts and webhooks
- Anonymous telemetry is on by default in release builds and is switched off with an environment variable
This is third-party code. Review the repository files before installing.
What it does
Kangentic is a development environment for coding agents in the form of a board. Drag a card to a column and Kangentic creates a git worktree, applies the column's permission mode and starts the agent. The card shows the model, how much of the context window is used and the agent's last message, and the ones that need an answer are marked. Fifteen agent CLIs are supported: Claude Code, Codex, Gemini CLI, Cursor CLI, GitHub Copilot CLI, Grok Build, Antigravity, Goose, OpenCode, Aider, Qwen Code, Kimi Code, Factory Droid, Warp and Ollama for local models. It runs the CLIs and keys you already have configured: no proxy and no Kangentic account. Columns are configurable: each has its own agent, model, effort and permission mode, and the setup lives in `kangentic.json` and travels with the repository. Automations on card entry, end of turn or exit can send the agent a message, run a script, call a webhook or move the card. Kangentic's MCP server lets a running agent create tasks, move its own card, search past conversations and message another agent. A mobile app shows agents and sends diffs with end-to-end encryption.
Who it is for. Developers who run several agents in parallel on different branches and lose track of which one is waiting for an answer.
Good fit when
- Five tasks run at once and one board should show which agent finished and which waits for you
- You want to plan in Claude Code and build in Codex, with the setup kept in the repository
- Card moves between columns should trigger scripts, webhooks or commands to an agent
- Tasks across several repositories should be tracked in one window with an agent monitor
- You want to answer a stalled agent from your phone and view its diff away from the desk
Not a fit when
- You prefer to work one task at a time, start to finish: the board pays off with several agents
- Agents must run in the cloud: Kangentic runs on your computer
- You use apps such as Cursor or Claude Desktop without a CLI: the board needs console agents
- You need a licence without copyleft: the project is AGPL-3.0, with a commercial licence address in the README
Example request
Create a Kangentic board task to fix the failing parser test, run it with the Claude Code agent in a separate worktree and tell me when the agent needs my answer.Limitations
Windows x64, macOS 13 or newer or Linux (deb and rpm), Node.js 20 for `npx`, Git 2.26 and at least one installed and signed-in agent CLI are required. The Knowledge Graph is unavailable on Intel Macs. On WSL the Windows version is installed, and a CLI installed only inside the distribution is not found.
Release builds send anonymous usage statistics (Aptabase) and error reports (Sentry) by default. Task text, prompts and code never go into them. The `KANGENTIC_TELEMETRY=0` variable turns them off. The mobile app connects through Kangentic's relay or your own copy, with end-to-end encrypted traffic.
How to disable. Quit the app and remove it the usual way for your system. Delete `kangentic.json` in the repository by hand; tasks and agent sessions live on your machine.
FAQ
How do I install Kangentic?
Run `npx kangentic`: it downloads the release for your system from GitHub Releases, installs it and opens it. After that the app updates itself. Installers are also on the releases page.
Do I need an account or an API key?
No. Kangentic launches the agent CLIs you have installed with the credentials already configured, with no proxy and no account of its own.
What can the Kangentic MCP server do?
A running agent creates tasks, moves its own card, searches past conversations and messages another agent. A rule in a prompt such as moving the card to review after tests then moves the work on its own.
Which agents are supported?
Fifteen: Claude Code, Codex, Gemini CLI, Cursor CLI, GitHub Copilot CLI, Grok Build, Antigravity CLI, Goose, OpenCode, Aider, Qwen Code, Kimi Code, Factory Droid, Warp (Oz) and Ollama for local models as single-turn chat. Resuming and the board's MCP tools depend on each CLI's capabilities.
Is the program paid?
No, per the README it is free and open under AGPL-3.0, with no paid tier and every feature included. For a commercial licence write to licensing@kangentic.com.
Discussion
No questions or reviews yet.
Related
CLI toolsOne desktop window for Claude Code, Codex, Gemini CLI and OpenCode: terminals, pull requests, diffs and approvals
Kanban from Cline launches a local web board where each card gets its own terminal and git worktree for a CLI agent.
A terminal coding agent in Go: GLM, DeepSeek and local Ollama models, 59 built-in tools and MCP
A Python terminal coding agent in the manner of Claude Code: reads the repository, edits files, runs commands, with memory, sub-agents and MCP
Agentic OS adds a five-step workflow and checks in git hooks and CI so that a step without evidence does not count as done
Kode CLI is a terminal coding agent that reads AGENTS.md and CLAUDE.md, connects MCP servers and works with many models.
Comparison with similar tools
Cards that share the main search query, and how they differ.
README in short
The README explains the problem of parallel agents and describes the board, mobile app, 15 CLIs, column setup, automations, workflows and the MCP server. Requirements, telemetry and interface languages are listed separately. Full documentation is on the project site.