User Manual

1. Welcome

This is the KaKeKiKoKu user manual — the complete reference for everything KaKeKiKoKu can do.

KaKeKiKoKu Studio is built for AI coding agents — Claude Code, Codex, Antigravity, OpenCode and the like — and for the projects you build with them, code or not.

KaKeKiKoKu Studio is the application: a knowledge system where you and your AI shape your project’s knowledge, kept structured, connected, and true to your files. It stores that knowledge in the Wuniq format, using files with the .wuniq extension.

If you just arrived

If this is your first time with KaKeKiKoKu, start with the Quick Start Guide on the KaKeKiKoKu website:

kakekikoku.com/quickstart

There you’ll find how to install and open Studio, choose a working directory, and connect your local AI with one instruction file. The common connection guide recommends a shared AGENTS.md, including for recent Claude Code versions, and explains how to keep existing Claude instructions or use your client’s equivalent. For everyday work, we recommend MCP; the MCP guides by client cover its setup.

The interactive tutorial

KaKeKiKoKu ships with the KaKeKiKoKu Studio Learning Lab: an interactive tutorial that is itself a real KaKeKiKoKu project, available in the six supported languages. When you set your working directory, KaKeKiKoKu offers to place it there — one click and it opens, already visible to your AI. Ten short stops teach the first mile of KaKeKiKoKu (context next to files, folders with rules, entities, relations, synchronization, working with AI), plus a practice playground with guided exercises. If you skipped the offer, use the Interactive tutorial link on the welcome screen.

The Learning Lab teaches you to walk; this manual is where you come for depth.

The welcome screen

The welcome screen is where you choose your working directory and decide which project folders belong to your KaKeKiKoKu workspace.

• Base working directory: the base folder where your AI client works. It usually contains several project folders, and KaKeKiKoKu requires it before you can open anything.
• Add project folder to workspace: browse for an empty folder, or an existing one with code, writing, research, or any other work. KaKeKiKoKu respects the files you already have.
• Add: adds the folder to the workspace list and marks it by default. Adding a folder does not open it yet.
• Open: use the Open button on a row to open one project, or Open marked projects to open every checked project together.
• Descriptions: if a folder has not been initialized as a KaKeKiKoKu project yet, KaKeKiKoKu asks for a short project description when you open it. That description helps your AI understand what the project is.
• Checks and order: the check decides which projects your AI can see and work with — whether they are open or not. Checked projects are the ones your AI can discover, and the ones opened by Open marked projects. An unchecked project can remain open for you without being offered through ordinary AI discovery, unless an AI has already opened it in the running application. You can reorder the list with the arrow buttons; unchecking a project keeps it in the list, and removing it from the list never deletes the folder from disk.
• Visibility is not access control: an AI-opened project remains visible across conversations connected to the running application. Unchecking does not revoke that state or erase what an AI already knows. Other tools may still access the files. Mark a project when you deliberately want the AI to discover it, then ask it to run open list.

If you already have KaKeKiKoKu running

This manual covers in detail everything you need to know:

• What is KaKeKiKoKu — the concept and why it exists (chapter 2).
• What KaKeKiKoKu is not — to avoid confusion with other tools (chapter 3).
• Core concepts — contexts, entities, sync, relation arrows, project map, and the three levels of detail (chapter 4).
• The interface — menu, navigation, editor, relationship arrows, consulting linked knowledge, project map, file and document viewers, and sync panel (chapters 5 to 8).
• Working with AI — practical tips, models, supervision, corrupt context recovery, and keeping a separate notebook out of ordinary AI discovery (chapter 9).
• Configuration — application and project settings (chapter 10).
• Troubleshooting — if something doesn't work, including corrupt contexts and platform limits (chapter 11).
• Tips and best practices — to get the most out of KaKeKiKoKu (chapter 12).
• Glossary — quick reference of terms (chapter 13).
• The knowledge file format — the plain-text format explained: read it anywhere, audit what KaKeKiKoKu writes into your repository (chapter 14).
• The KaKeKiKoKu AI Guide — the introductory help and every addressable help topic your AI can retrieve from KaKeKiKoKu, published verbatim. The format chapter covers what KaKeKiKoKu writes; this covers what your AI is told (annex).

Tip

You don't need to read the entire manual. Start using KaKeKiKoKu and consult the chapters you need. The contextual helps within the application will guide you through each feature the first time you use it.

Need help? Pass the URL of this documentation to your AI and ask it to read it. Your AI is your best guide for resolving questions about KaKeKiKoKu.