Kagan
Drive autonomous dev work from a terminal Kanban board while MCP-connected agents pick up and execute cards.
Overview
Kagan is a MCP server for the Build phase that exposes an AI-powered Kanban TUI and orchestrates autonomous development workflows through stdio MCP.
What is this MCP server?
- AI-powered Kanban TUI with MCP stdio transport (uvx / PyPI package kagan)
- Agent execution via KAGAN_OPENAI_API_KEY or KAGAN_ANTHROPIC_API_KEY
- Persistent workflow state with optional KAGAN_DB_PATH for the Kagan database
- Autonomous development workflow orchestration from terminal UI plus MCP
- Beta server release 0.11.2-beta.1 from kagan-sh/kagan on GitHub
- Server version 0.11.2-beta.1
- stdio transport via PyPI identifier kagan
- 2 optional LLM provider secret env vars plus KAGAN_DB_PATH
Community signal: 96 GitHub stars.
What problem does it solve?
Coding agents drift when work lives only in chat logs with no shared board, state, or repeatable execution path.
Who is it for?
Solo builders running Claude or OpenAI-backed agents who want a terminal Kanban control plane wired into MCP.
Skip if: Teams that only need a human-only Kanban with no agent backends or MCP integration.
What do I get? / Deliverables
After you register Kagan, agents and you share a Kanban-backed queue with optional DB persistence and API-key-driven execution.
- MCP-connected Kanban workflow for agent tasks
- Optional on-disk board state via KAGAN_DB_PATH
- Orchestrated agent runs tied to board cards
Recommended MCP Servers
Journey fit
Kagan is where solo builders wire agent orchestration into day-to-day implementation, not a one-off script. Kanban plus agent backends maps directly to agent-tooling: tasks, runs, and orchestration from the TUI/MCP surface.
How it compares
Agent-orchestrated Kanban MCP server, not a generic project-management SaaS skill.
Common Questions / FAQ
Who is Kagan for?
Kagan is for indie developers and agent-heavy workflows that want a TUI Kanban board connected to OpenAI or Anthropic execution over MCP.
When should I use Kagan?
Use Kagan when you are building or iterating with autonomous agents and need visible columns, persisted tasks, and MCP-accessible orchestration.
How do I add Kagan to my agent?
Add the stdio MCP entry for package kagan (uvx from PyPI), set KAGAN_OPENAI_API_KEY or KAGAN_ANTHROPIC_API_KEY, optionally KAGAN_DB_PATH, then restart or reload MCP in Claude Code, Cursor, or Codex.