May 24, 2026

KaKeKiKoKu Studio 1.4: images as knowledge

A small release about a simple idea: some images deserve memory.

I am still iterating on Studio little by little, faithful to the way this project has been built: small releases, concrete improvements, and no sudden attempt to become a different product.

KaKeKiKoKu Studio 1.4 follows that line.

The main change is simple: an image no longer has to be just an image.

Almost every real project ends up with visual material somewhere. Screenshots. Mockups. Diagrams. References. Covers. Maps. A design that settled an argument faster than ten paragraphs ever could.

But an image by itself can still be silent.

It can show an interface without explaining the decision behind it. It can capture a design without remembering the alternative that was rejected. It can be visually obvious to you today and completely opaque to an AI that returns to the project next week.

That is the small gap KaKeKiKoKu Studio 1.4 closes.

Images with context

In KaKeKiKoKu Studio 1.4, an important image can have its own context card. You can open the image, see it large, and keep the notes that explain what it represents right beside it.

What does this image mean? Why is it here? Which decision does it preserve? What should the AI remember when it sees this part of the project again?

Those questions are exactly the kind of knowledge Studio was built to preserve.

This is not an asset manager. Studio does not take over your files, import them into a separate system, or ask you to reorganize the project around it. The image stays where you put it. Studio adds the knowledge layer around it.

That distinction matters to me. Studio should not become a storage product. It should remain what it is: a local Knowledge Engine that lives beside your work and makes the project more understandable for you and for the AI working with you.

Why it matters

Knowledge is not only text.

A screenshot can explain a bug. A mockup can freeze a product decision. A diagram can carry architectural intent. A reference image can say something about tone, shape, or direction that would become clumsy if you tried to translate it into prose.

Before 1.4, Studio could preserve the notes around those things, but the visual piece itself stayed a little outside the experience. Now the image can sit inside the same mental model as any other important resource: it can be opened, described, connected, and understood as part of the project memory.

Relations to images also become easier to recognize. When a documented image is linked from another resource, Studio can show a thumbnail instead of a generic link icon. It is a small visual change, but it makes the graph feel less abstract. You recognize the thing before you read the path.

A quieter viewer

KaKeKiKoKu Studio 1.4 also improves the central viewer.

There is now a Back button in the breadcrumbs, so you can return to the resource you were viewing before. The header is clearer, with navigation separated from status and actions. Some technical information that only the AI needed has been removed from the human view.

None of that is glamorous. It is just the kind of polishing that makes a tool feel less tiring after a long session.

There is also a small but important fix around closing projects: when you close a project in Studio, future AI sessions will no longer see it as open. That is the kind of detail that matters when the tool is not a toy demo, but part of a real working environment.

Where this points

This release also points toward the next shape of Studio.

Not as a public checklist of promises, but as a direction: relations should become easier to create and understand, the central viewer should show richer context around each resource, and writing visual or Markdown-based knowledge should become less manual over time.

That is the roadmap in its honest form. Not a grand leap. A sequence of small steps around the same idea.

Your AI should not come back to the project blind.

With 1.4, the images that matter can become part of that memory too.