Kagan

Kagan is an OpenCode plugin that turns AI coding agent work into supervised kanban tasks, each running in an isolated git worktree with review gates.

What you can do

Kagan presents a kanban board (Backlog, In Progress, Review, Done) where each task is an OpenCode session running in its own git worktree on a kagan/ branch. A reviewer agent files ranked findings against the original task, and nothing reaches Done until the user triages every finding and decides where or whether to merge. Tasks can be created conversationally with /kagan-task, and the board is opened with /kagan or a leader-key shortcut.

When to use it

Use it when you want AI coding agents to work on multiple tickets under supervision, with review checkpoints before anything is merged. It suits developers already using OpenCode who want agent work isolated from their checkout and gated at each transition.

Requirements

Runs locally as an OpenCode plugin; OpenCode 1.17.13 or newer is required. Installed globally via the OpenCode plugin command or from a local clone added to OpenCode config. Agent execution needs an Anthropic API key (KAGAN_ANTHROPIC_API_KEY) or an OpenAI API key (KAGAN_OPENAI_API_KEY) depending on the backend. KAGAN_DB_PATH optionally sets a custom database file path.