kenn-io/kata

What it solves

Kata is a local-first issue tracker designed specifically for the collaboration between humans and AI coding agents. It replaces chat threads and markdown lists with a durable, structured task ledger that agents can programmatically operate via CLI, while humans can supervise the work via a terminal UI (TUI) or browser-based UI.

How it works

Kata operates as a local service that manages issue state in a SQLite database (or Postgres for remote teams) rather than storing issues within the git repository itself. This keeps the repository clean. It provides a CLI for creating, claiming, and closing issues with evidence, and integrates with the Model Context Protocol (MCP) to allow AI agents to interact with the task ledger directly.

Who it’s for

Developers and AI agent harnesses (like Claude Code or Codex) that need a structured way to track progress, assign tasks, and verify completions without relying on a heavy SaaS project management tool.

Highlights

• Agent-First Ergonomics: Provides stable short references, idempotent creates, and specialized --agent output for predictable AI scripting.
• Local-First Architecture: Uses a single Go binary with no runtime dependencies, keeping issue history separate from code history.
• Evidence-Based Closing: Requires explicit completion claims with reasons, messages, and evidence (like commit SHAs) for auditable work.
• MCP Integration: Native server support for MCP clients to expose typed issue and automation workflows.
• Flexible Deployment: Supports local SQLite by default, but can scale to a remote daemon or federation using PostgreSQL.