[PR #492] [CLOSED] Add Kagan - AI-Powered Kanban for Multi-Agent Autonomous Development #2924

New Issue

GiteaMirror · 2026-04-26T21:58:10-05:00

GiteaMirror commented

2026-04-26 21:58:10 -05:00

📋 Pull Request Information

Original PR: https://github.com/Shubhamsaboo/awesome-llm-apps/pull/492
Author: @aorumbayev
Created: 2/12/2026
Status: ❌ Closed

Base: main ← Head: add-kagan

📝 Commits (3)

176a570 Add Kagan to Multi-agent Teams section
3ceed97 Add Kagan multi-agent development tutorial
67de4ea fix: update Kagan link to point to subfolder instead of external URL

📊 Changes

2 files changed (+206 additions, -0 deletions)

View changed files

📝 README.md (+1 -0)
➕ advanced_ai_agents/multi_agent_apps/agent_teams/kagan_multi_agent_dev/README.md (+205 -0)

📄 Description

Summary

Adds Kagan to the Multi-agent Teams section.

About Kagan

Kagan is a terminal Kanban board + MCP server for autonomous development. Developers manage tasks, pair-program with AI agents, and launch IDE coding sessions — all from a TUI. The standalone core daemon exposes a full MCP interface, turning Kagan into an orchestration environment where an external AI CLI (with admin-role MCP) can delegate tasks across completely different coding agents.

Two Interfaces, One Workflow

TUI (for developers): Interactive Kanban board where you create tasks, initiate AUTO (background) or PAIR (interactive) coding sessions, launch agents directly into tmux/VS Code/Cursor, review diffs, approve, and merge — without leaving the terminal.

MCP Server (for AI CLIs): An admin-role MCP client (Claude Code, OpenCode, Codex, etc.) connects to Kagan and uses it as a persistent orchestration layer — creating tasks, dispatching them to different AI coding agents, tracking progress with structured logs, and coordinating reviews. This means Claude Code can delegate a task to Gemini CLI, monitor it through Kagan, then assign the next task to Kimi CLI — all with persistent state, logging, and audit trail.

Key Features

6 built-in AI coding agents: Claude Code, OpenCode (SST), Codex (OpenAI), Gemini CLI (Google), Kimi CLI (Moonshot AI), GitHub Copilot
Cross-agent orchestration: An admin MCP client can delegate tasks to any supported agent — connect Claude with Gemini with Kimi with Copilot through a single Kanban
MCP server: task CRUD, job automation, code review, merge/rebase, session management, settings, audit
Kanban workflow: BACKLOG → IN_PROGRESS → REVIEW → DONE with built-in review/approve/reject/merge
IDE session launching: Initiate PAIR sessions directly into tmux, VS Code, or Cursor from the TUI
Persistent logging: Every agent run is tracked with structured execution logs, scratchpad notes, and audit events
Git worktree isolation: Each task gets its own branch and worktree — parallel work without conflicts
5 MCP capability profiles: viewer → planner → pair_worker → operator → maintainer for fine-grained access control
Chat-driven planning: AI-powered task decomposition with approval flow

Architecture

Developer ──► TUI (Kanban Board)
                    │
                    ▼
              Core Daemon ◄── MCP Server ◄── External AI CLI (admin role)
                    │
          ┌─────────┼─────────┐
          ▼         ▼         ▼
     Claude Code  Gemini   Kimi CLI  ...

Tech Stack: Python, Textual, FastMCP, SQLModel, Click, Rich

Repo: https://github.com/kagan-sh/kagan
Docs: https://docs.kagan.sh
PyPI: https://pypi.org/project/kagan/

_{🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.}

## 📋 Pull Request Information **Original PR:** https://github.com/Shubhamsaboo/awesome-llm-apps/pull/492 **Author:** [@aorumbayev](https://github.com/aorumbayev) **Created:** 2/12/2026 **Status:** ❌ Closed **Base:** `main` ← **Head:** `add-kagan` --- ### 📝 Commits (3) - [`176a570`](https://github.com/Shubhamsaboo/awesome-llm-apps/commit/176a570c2a8af96def91607496b977b179443c48) Add Kagan to Multi-agent Teams section - [`3ceed97`](https://github.com/Shubhamsaboo/awesome-llm-apps/commit/3ceed971109f8606e10c11f69aecfe98ec01e524) Add Kagan multi-agent development tutorial - [`67de4ea`](https://github.com/Shubhamsaboo/awesome-llm-apps/commit/67de4ea1a08fdd24cf3d2eeeaf9a88e3d45b10b9) fix: update Kagan link to point to subfolder instead of external URL ### 📊 Changes **2 files changed** (+206 additions, -0 deletions) <details> <summary>View changed files</summary> 📝 `README.md` (+1 -0) ➕ `advanced_ai_agents/multi_agent_apps/agent_teams/kagan_multi_agent_dev/README.md` (+205 -0) </details> ### 📄 Description ## Summary Adds [Kagan](https://github.com/kagan-sh/kagan) to the Multi-agent Teams section. ## About Kagan <p align="center"> <img src="https://raw.githubusercontent.com/kagan-sh/kagan/main/.github/assets/demo.gif" alt="Kagan Demo" width="700"> </p> Kagan is a **terminal Kanban board + MCP server** for autonomous development. Developers manage tasks, pair-program with AI agents, and launch IDE coding sessions — all from a TUI. The standalone core daemon exposes a full MCP interface, turning Kagan into an **orchestration environment** where an external AI CLI (with admin-role MCP) can delegate tasks across completely different coding agents. ### Two Interfaces, One Workflow **TUI (for developers):** Interactive Kanban board where you create tasks, initiate AUTO (background) or PAIR (interactive) coding sessions, launch agents directly into tmux/VS Code/Cursor, review diffs, approve, and merge — without leaving the terminal. **MCP Server (for AI CLIs):** An admin-role MCP client (Claude Code, OpenCode, Codex, etc.) connects to Kagan and uses it as a persistent orchestration layer — creating tasks, dispatching them to *different* AI coding agents, tracking progress with structured logs, and coordinating reviews. This means Claude Code can delegate a task to Gemini CLI, monitor it through Kagan, then assign the next task to Kimi CLI — all with persistent state, logging, and audit trail. ### Key Features - **6 built-in AI coding agents**: Claude Code, OpenCode (SST), Codex (OpenAI), Gemini CLI (Google), Kimi CLI (Moonshot AI), GitHub Copilot - **Cross-agent orchestration**: An admin MCP client can delegate tasks to any supported agent — connect Claude with Gemini with Kimi with Copilot through a single Kanban - **MCP server**: task CRUD, job automation, code review, merge/rebase, session management, settings, audit - **Kanban workflow**: BACKLOG → IN_PROGRESS → REVIEW → DONE with built-in review/approve/reject/merge - **IDE session launching**: Initiate PAIR sessions directly into tmux, VS Code, or Cursor from the TUI - **Persistent logging**: Every agent run is tracked with structured execution logs, scratchpad notes, and audit events - **Git worktree isolation**: Each task gets its own branch and worktree — parallel work without conflicts - **5 MCP capability profiles**: viewer → planner → pair_worker → operator → maintainer for fine-grained access control - **Chat-driven planning**: AI-powered task decomposition with approval flow ### Architecture ``` Developer ──► TUI (Kanban Board) │ ▼ Core Daemon ◄── MCP Server ◄── External AI CLI (admin role) │ ┌─────────┼─────────┐ ▼ ▼ ▼ Claude Code Gemini Kimi CLI ... ``` **Tech Stack:** Python, Textual, FastMCP, SQLModel, Click, Rich **Repo:** https://github.com/kagan-sh/kagan **Docs:** https://docs.kagan.sh **PyPI:** https://pypi.org/project/kagan/ --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

GiteaMirror added the pull-request label 2026-04-26 21:58:10 -05:00

GiteaMirror closed this issue

2026-04-26 21:58:11 -05:00

Sign in to join this conversation.

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: github-starred/awesome-llm-apps#2924