The Museum of All Things is a 3D gallery that Wikipedia builds for you

The museum is generated from the link graph

The goal of the project is stated plainly, to make an interactive 3D museum that is generated procedurally, using content from Wikipedia to fill exhibits, and the museum is described as virtually limitless because you take doors from one exhibit to another depending on what is linked from the Wikipedia article you are standing in front of. That is the whole idea in one sentence, and it is a better idea than it first sounds, because the ordering of the rooms is decided by the encyclopedia rather than by a curator, which means the sequence of things you see is a kind of associative walk through the links between subjects. Every exhibit is filled with hallways to other exhibits based on the links on the current page, so the claim in the readme is that you will never run out of things to explore. The text of the article is inserted as informative plaques on the wall, so you can read about the exhibit while looking at the pictures from it, and the images are pulled from Wikimedia Commons in the category corresponding to the article. The design therefore has three content sources, the article for the words, Commons for the pictures, and the link structure for the floor plan.

Godot, a VR build and a lot of asset directories

The game is written in the Godot engine, and the primary language of the repository is GDScript, which is Godot's own language. The tree shows the shape of a real project rather than a sketch. There is a project file, a scenes directory, a shaders directory, a resource definitions directory, an addons directory for plugins, a Linux directory, an export presets configuration, a default audio bus layout, and an OpenXR action map together with a VR splash screen, which means virtual reality is a supported target rather than an aspiration. There is also a Blender directory, which is where the source models live, and an itch.io manifest, so the project is published as a downloadable build as well as through GitHub releases. The project page is hosted separately from the repository, and the readme links both the GitHub releases and the itch page. The distribution story is therefore two channels, a code repository for people who want to build it, and store pages for people who want to play it.

The contribution rule bans generated code, which is unusual

The contributing section opens with a line in capital letters, that any code or assets contributed must be written by human hands, and that generative AI output is not accepted into the project. In a project whose entire premise is procedural generation from a vast external corpus, that is a deliberate and slightly surprising line to draw, and it is worth reading as a statement about authorship rather than about technology. The practical consequence is that pull requests containing generated code or generated art will be rejected, whatever their quality. The same section is explicit about the second constraint, which is about scope. The author is a solo developer distributing a free project, will prioritise whatever they have time for and feel motivated to work on, and is more likely to accept pull requests that are bug fixes, optimisations or visual upgrades, while being precious about the creative direction of the museum, so those changes come with no guarantee of being merged. The advice given is to file an issue and discuss before doing the work, which is the only way to avoid a pull request that is never merged.

Localisation is where the community is

The readme is candid that it does not currently have a policy for allowing outside contributors to the codebase, and that this might change if the author pursues localisation in other languages. But that is immediately followed by an invitation, with a link to a translation guide asking whether you are interested in adding your language to the museum, and a statement that there are now many contributors and that pull requests will be reviewed. The list of supported languages on the main branch, which may not reflect the latest release, runs to eight: English, Portuguese, French, Spanish, Japanese, German, Bengali and Chinese. The roadmap records that support for more Wikipedia languages other than English is done, with thanks to a named contributor and others. So the project has found a way to open up contribution without opening up the design, and localisation is the seam. For a reader, that also means the museum's coverage depends on which language editions of Wikipedia are wired in, which is a constraint on the procedural generation itself rather than on the interface.

The roadmap is honest about the big item

The future features list has four entries, and the first one is multiplayer, annotated with the note that this one is big and not to be expected anytime soon. The second is greater variety in theming and exhibit layouts, the third is support for more media types such as audio or 3D models, and the fourth, support for more Wikipedia languages other than English, is struck through as done. The annotation on multiplayer is the interesting part, because it is the feature that would change the project from a solo wander into a social space, and the author says plainly that it is a big one. The struck through entry shows the other half of the pattern, which is that the project does complete things it sets out to do. The combination of a long standing big item, a clear list of smaller ones and an honest annotation is what a well run solo project looks like, and it is a better signal than a feature list would be.

Credits, licences and where the assets come from

The credits section separates the code from the content, which matters for a project that assembles other people's work. All exhibit content is sourced from Wikipedia and Wikimedia Commons, the game is written in the Godot engine, and the textures come from a separate asset site. The code itself is MIT licensed. The people are listed by role, with the creator and programmer credited first, then app store publishing and multi platform support, then audio with a note that the audio contributor is accepting work, and then dramaturgy. That last role is a good signal about what the project thinks it is, since a museum needs someone thinking about sequence and staging, not just geometry. The overall credit list also explains why the project has a VR build and a store presence at all, since those were contributions rather than the original plan, and it is a reasonable summary of the shape of the whole effort, a solo core with a community layer added on top.

Editorial conclusion

Museum of All Things earns its place for anyone who wants an endless, unhurried way to browse the encyclopedia, since the linking behaviour of Wikipedia produces a museum whose rooms are ordered by association rather than by an editor, and the article text on the wall means you can read while you look. It is the wrong choice if you want a curated tour, since the point is that the curator is the link graph, and the wrong choice if you want to extend the exhibition yourself, because the contribution policy is explicit that generated code and assets are not accepted. Verify three things before you judge it: that you are comfortable with the content arriving from Wikipedia at play time rather than from a curated set, since the museum is generated from live data, that the VR and desktop builds both work on the platform you have, since the repository carries OpenXR action maps and a VR splash screen alongside the standard export presets, and that the licence terms are compatible with what you do with it, since the code is MIT while the exhibit content belongs to Wikipedia and Wikimedia Commons and the textures come from a separate source. The project is MIT licensed, the last push was on 2026-10-15, and the newest release is v1.2.0 from 2026-06-06.

Frequently asked questions

Official sources

1. Issues
2. License: MIT
3. m4ym4y/museum-of-all-things on GitHub
4. README
5. Releases