Kagan - AI Orchestration Layer

Kagan is an AI orchestration layer providing a Kanban-style interface for managing coding tasks, automating development workflows with AI agents, and integrating with external systems.

Task Management — Create (individually or in batch), list, get, update, delete, search, and add notes to tasks. Tasks support properties like title, description, priority, base branch, acceptance criteria, and agent backend. Also view task events, wait for status changes, and get task counts.

Agent Run Management — Start, cancel, kill, and manage interactive sessions for AI agents working on tasks. Retrieve run status/summaries, check for active sessions, and detach from sessions.

Project & Repository Management — Create, list, delete, and activate projects. Add repositories to projects and configure default branches.

Code Review Workflow — Approve, reject (with feedback), and merge review-ready tasks. Perform Git operations like rebasing, conflict handling, and rebase progression. Set or clear AI review verdicts for individual acceptance criteria.

Settings & Audit — Retrieve and modify system configuration settings; access audit logs.

Persona Preset Management — Audit, import, and export AI persona presets from/to repositories; manage a whitelist of trusted persona sources.

Plugin Integrations — Sync external items (e.g., GitHub issues) into Kagan tasks, and preflight-check plugin dependencies.

Interfaces & Agent Modes — Interact via a terminal Kanban board (TUI) or web dashboard. Supports both autonomous and pair programming modes across 14 AI coding agents.

Provides comprehensive management of Git repositories through task-based branches, supporting operations such as rebasing, merging approved changes, and tracking repository-specific branches.

Enables synchronization of GitHub issues into the Kanban board, mapping labels like priority and status to task properties while providing tools to verify GitHub CLI authentication.