Recovered transmissions from the edge of development

The CosmOS website has become something a little stranger than a conventional game-development homepage. There is an AEGIS workstation to explore, an Arcade full of experiments, creative studios for making things of your own, and recovered records from across the universe. There are also a few reasons to pay attention to the things lying around.

For a while, we wrote individual release posts whenever a new Arcade experience arrived. Now the collection changes too quickly for those articles to capture the whole picture. This post is a tour of the Archives themselves—what's playable, what you can make, and what you might discover.

Inside AEGIS_OS

The Archives live in a working AEGIS-style desktop: windows, taskbar, Arcade, file explorer, media, and applications rather than a static list of releases. Launch the AEGIS_OS workstation →

Games to explore

Browse the current playable and prototype collection, from physics experiments and adventures to world-building and simulations. These are developing browser experiences, not the completed CosmOS VR campaign.

Studios and creative tools

Back in AEGIS_OS, you'll also find the desktop's Browser, Mail, Media Player, and other familiar workstation utilities—small pieces of the Archives' identity as a place you can explore.

Secrets are part of the invitation

Not everything on the website announces itself as a menu option. Objects and discoveries across the site can lead you toward additional Arcade experiences. Some doors open through curiosity: checking unusual objects, paying attention to what responds to interaction, or revisiting somewhere after learning something new.

We're intentionally not turning this into a complete solution guide. The fun is in noticing that an ordinary-looking page can contain something you can touch, move or uncover. The site companions bring some of that playful continuity along for the journey, with their own behaviours and activities to explore.

Some secrets are small; others point toward larger systems or places. Either way, exploration should feel rewarding even before you know what you're looking for.

How the Archives got here

The Archives began with art, comics, wallpapers and early glimpses of the universe. The May 3 retrospective, The Archives Take Shape, occupies an editorial chronology position; the later Archives reorganization article originally appeared on August 16, 2026. These were not launches of the Arcade games.

Playable experiments arrived separately: Voidway Sandbox was announced June 13, while Construct Command and Cutie Find-it were announced August 13. The Archives have since grown into a connected desktop of games, creative tools, stories and secrets.

Why we're consolidating updates

The older individual Arcade release posts will stay as snapshots of those experiments when they first appeared. But a new mechanic, an expanded studio or an improved desktop app doesn't always need a standalone announcement. Going forward, we want Archives updates to connect the dots: what you can play, what you can make, what changed, and what the pieces teach us.

We'll still mark particularly substantial releases on their own when it makes sense. The Archives will continue evolving between those milestones.

Open the Archives and explore →

Interaction & Persistence examined how an action might become a meaningful consequence. But before we ask the world to remember an action, we should ask what makes that action interesting to perform.

What makes a game complex? The obvious answer is to count its mechanics. A character can walk, sprint, jump, roll, slide or crawl; surely that is six rules?

We're exploring a different way to think about it. Count the systems that change the player's agency, not every expression of those systems.

One rule, many manifestations

Consider locomotion. Walking and rolling feel different. Jumping changes the relationship with the ground. Crawling changes the spaces a character can enter. But each is a manifestation of the same governing question: how can the player move through the world?

Now place that character on a steep hill, a narrow bridge, a curved ramp or inside a hollow log. The environment creates new demands and opportunities. We have not necessarily added a new rule; we have discovered more situations in which locomotion matters.

That is one reason a satisfying controller can be fun before there is a level or objective. A small, coherent ruleset can have a surprisingly large possibility space.

What counts as a rule?

Our working definition is deliberately practical: a rule is a governing system that changes what the player can do in the world. A manifestation is a mode of that system. An interaction is what happens when a mode meets the environment or another system.

That definition also changes how we think about rewards. A coin that only adds to a score measures performance, but does not, by itself, give the player new agency. If coins can be spent on abilities, tools or changes to the world, an economy has entered the design. The player's possibilities now depend on what they acquire and how they spend it.

Likewise, a checkpoint is more than a marker when it changes where the player returns after failure. A movable crate might be a simple physical expression of existing movement and collision, or part of a broader manipulation system. The distinction rests on the capabilities the game actually gives the player, not the name we assign to a mechanic.

Why the count matters

During a small development challenge, we found ourselves comparing two experiments. One was a browser adventure built around handcrafted scenes, guards, keys and story interactions. The other was a third-person Unity controller built for a playful mix of walking and physical rolling.

The browser adventure has been harder to make satisfying: each specialized interaction demands behaviour, exceptions and testing. The controller, by contrast, is already enjoyable simply to move around with. That is an observation about these prototypes, not a universal verdict on either genre.

It suggests a useful development discipline: prove the pleasure of one rule before surrounding it with many others. A level can explore that rule through shape, scale, friction, elevation and timing without becoming a catalogue of one-off mechanics.

When a second rule earns its place

Not every addition is unnecessary. A new rule can multiply what existing rules make possible. Wind, for example, could affect a rolling character on slopes, a jump across a gap and the trajectory of an airborne landing. One system creates consequences across many situations.

Compare that with a button whose only purpose is to open a single unrelated door. It can still serve a scene or a story, but it offers less systemic reach. Neither is automatically wrong; they have different design costs and different rewards.

We are not proposing a literal formula that proves one game better than another. Rule boundaries are partly a design choice, and a narrowly scripted interaction may be exactly what a particular story needs. The count is a lens, not a law of nature.

Build possibility, not just complexity

Before adding a system, the questions become: What can the player do now that they could not do before? Does this deepen an existing rule, or introduce a new one? And how many meaningful interactions does that addition create?

For The CosmOS, this returns us to the question of interaction: worlds should respond to what players do, rather than merely to what they select. A strong foundation should support many expressions of agency without requiring every meaningful moment to be separately scripted.

The goal is not the fewest rules at any cost. It is to make each rule worth having.

The CosmOS Site Companion area is now live: a new way to keep a little company while exploring the website.

Choose an active pet from the site's companion picker and take it with you as you move between pages. Your selection and companion state carry across ordinary site navigation. You can also select No pet whenever you want a clear screen.

From page to page, even into the Arcade

On ordinary pages, companions use the site's draggable-object physics. They respond to surfaces and loose discoveries rather than simply following the cursor, while the companion picker lets you change which pet is active.

The same selection can accompany you into supported Archives games. Look for CosmOS Site Companion beside the game's existing options and turn it on to show your pet in a separate overlay. It starts switched off, so joining a game remains optional and does not change the game's own mechanics.

A closer look at the Critter Construct

Critter explores nearby controls, faces the direction he is moving and rolls through the page inside his holographic shell. He can gently nudge loose objects, react when something catches his attention and slow down if you keep the pointer over his ball long enough to pet him.

His movement and changing expressions give this first Construct companion a personality of his own. Here is a selection from his production sprite atlas.

Idle

Rolling

Curious

Affection

Sleeping

The companion system is a contained experiment in continuity and presence: what changes when something on a website notices a visitor, responds to touch and stays with them beyond a single page?

Zub: discover the rest yourself

The egg is only the beginning. Zub develops through focused visits and hands-on care. If you would rather find out what happens by looking after him, leave the optional section below closed.

Reveal Zub's growth and interactions (evolution spoilers)

Each growing stage requires ten minutes of focused site time and ten deliberate care interactions before Zub evolves. Background or offline time does not count towards that progress.

Egg

Larva

Beetle

Winged adult

What changes as Zub grows

Egg

The shell changes and gradually cracks as its focused-time progress advances. Even before hatching, a deliberate hover counts as caring for Zub.

Larva

Zub begins to crawl and respond to petting. He can encounter edible loose objects, or you can hold a suitable collectible against him to hand-feed him.

Beetle

With a more developed walking form, Zub keeps the physical pickup, petting and feeding interactions while working towards his final evolution.

Winged adult

Zub now alternates between resting, preparing for take-off, hovering and landing. Pick him up and throw him and he can recover into flight; his new form still responds to petting and edible discoveries.

Care is an active choice: a completed hover counts once per entry, while deliberate hand-feeding suitable food can contribute care during growing stages. Reached forms become selectable without resetting the live growing companion.

October 16, 2026 · Part I is now open

Our first installment of The First Threshold is ready to read. Part I — Arrival introduces Natalie Vale, a young woman beginning a new job at Aegis in San Francisco. Between an unfamiliar city, a demanding office and the people she meets along the way, Natalie is finding her footing in a world that still seems entirely ordinary.

It won't remain ordinary forever.

The First Threshold tells a story from the years before The CosmOS became the universe we know. It begins with people and the decisions they make, before the discoveries that will change their lives. Part I brings together the first three chapters: Arrival, The Call, and A Less Impossible Day.

Read The First Threshold — Part I: Arrival →

This is an opening installment, not the complete novel. The reader's contents also show the remaining Parts — Launch Day, The Golden Age, The Collapse, and The Last Greeter — as locked chapters of the journey still to come.

September 26, 2026 · Building a place to read

The CosmOS is being built to hold more than playable worlds. Some parts of its history belong in longer-form stories, and those stories deserve a reading experience of their own.

We designed a focused long-form reader around that idea: expandable contents, a reading-progress indicator, adjustable text size, paper and night themes, and a reading position that is remembered between visits. The aim is to make room for the story without asking readers to learn a complicated interface.

Our original interactive reader walkthrough remains available with synthetic demonstration pages. It documents how the reading experience took shape, while the Archives edition is now the place to begin the real story.

Thank you for reading. We hope you'll enjoy meeting Natalie.

How actions become part of a world.

Foundations began with a promise: to examine the systems beneath The CosmOS, one subject at a time, and distinguish what works from what we hope to build. We begin where the player meets the world.

A world is not made believable by the number of things it contains. What matters is what happens when someone reaches into it. Does an object respond as expected? Can a player understand the result? And if an action changes something important, does the world remember?

These questions connect several areas of The CosmOS that were previously described in separate design notes. Player presence, material behaviour, actors, readable power and world consequence are not five unrelated promises. They are different parts of one design problem: how do we let a player act within a world that responds coherently and can carry those actions forward?

This entry maps the intended relationship between those areas; it does not announce a complete, playable interaction and persistence architecture. The distinction matters: we can explain what the systems must accomplish without claiming they are finished.

The first point of contact

Interaction begins with the player. In The CosmOS, the ambition for presence goes beyond moving a viewpoint through scenery. The player's body, position and contact with the environment should matter. Water, altered gravity, hazards and other physical conditions raise a common question: how can the player tell what is happening, and respond deliberately?

This is the player-side foundation of interaction. Without understandable feedback, physical behaviour becomes noise; without a coherent body, a world can feel like a collection of disconnected controls.

Our earlier player-presence note recorded these questions and a wider set of design considerations. It should not be read as evidence that every condition mentioned there was integrated or publicly available at the time. Working implementations deserve their own verified Foundations articles, with demonstrations and limitations rather than a retrospective feature checklist.

A language shared by objects and actors

The next challenge is the things the player meets. Different objects and characters need legible ways to expose their state and respond to an action. The details can vary, but scene authors and players should be able to understand the points of contact: what can be acted on, what conditions apply, what changed and what might happen next.

The architectural ambition is to favour reusable interaction behaviour over bespoke rules hidden inside every scene. Exactly how those boundaries are packaged and made available remains separate engineering work; this article is about the design relationship, not an SDK release announcement.

Rules the world can teach

Materials and power introduce another kind of interaction: the environment itself can explain what its rules allow.

The earlier material-grammar note asked whether contact, heat, moisture, pressure, charge and containment could form an understandable vocabulary. Its examples were candidate rules to develop and test, not a list of confirmed reactions in the public Voidway Field. The Arcade sandbox is a separate, contained browser experiment with its own actual feature set.

The readable-power note explored a related question through two distinct concepts: performed glyphs and environmental node circuits. These are not interchangeable systems or announced player abilities. Their common design goal is readability. Ideally, a player should understand the relationship between an action and its outcome, rather than seeing an unexplained effect triggered by a menu.

This is the bridge from interaction to agency: players need enough consistency to form expectations, experiment, make mistakes and learn.

When a result becomes history

A world that responds in the moment can still feel temporary. The larger CosmOS ambition asks what happens after the immediate reaction.

Imagine a route changed by an action or a structure restored. A meaningful consequence is not simply an animation or a journal message. The changed state needs to be represented somewhere, and a later experience needs a reliable way to interpret it. That is the purpose of persistence: to carry selected, meaningful changes beyond the instant in which they occurred.

There are difficult design questions beneath that promise. Which changes deserve to be remembered? What remains local to a player, a scene or a world? How should a later visit interpret a past event without turning every encounter into a special-case script? And what does it mean to restore a place if the world cannot distinguish the damaged state from the repaired one?

Renown, the Player Chronicle and world-state systems may all contribute to the wider design. Their mention here is not a claim that a universal persistence simulation, autonomous actor memory or every proposed consequence has shipped. Those claims require implementation evidence and their own dedicated updates.

One chain, many worlds

The intended relationship is straightforward, even when the engineering is not:

A player acts. Something receives the action. The world produces a readable response. Important changes become state that later experiences can use.

The same chain can support a tiny interactive scene or a sprawling authored world. It is also why we cannot treat interaction, material behaviour, actors and persistence as isolated feature checkboxes. Their value grows when their boundaries and responsibilities fit together.

This does not mean every object or footstep should persist forever. Selectivity is part of making a coherent experience. The challenge is to remember what is meaningful, express that memory through the world, and avoid claiming consequences that do not actually survive beyond the moment.

Showing the proof

There are concrete foundations beneath this direction. The public Arcade experiments already let us test bounded inputs and readable responses, while the project's established design records define how larger systems are intended to connect. Those are real pieces of the work, but they are not proof that the complete interaction-to-persistence chain is finished.

The early notes established a direction. Future Foundations entries can test it against what has actually been implemented, what failed and how each part connects to the wider experience. A materials experiment can be demonstrated on its own terms. An embodiment update can show real movement and contact. A later persistence entry can follow a single event through its state changes before making broader claims.

The CosmOS is intended to be more than a set of places to look at. It is meant to be a set of places where actions can matter. Before deciding what the world should remember, though, we need to understand what makes an action worth performing.

Interaction is how a player reaches into a world. Persistence is how the world reaches back later.

Every universe has something holding it together.

The CosmOS is an ambitious project: a universe of massive worlds, each with its own identity, history and possibilities.

Building that universe means working on far more than what appears on the surface.

Behind every environment, every interaction and every meaningful choice lies a foundation that needs to hold. Those foundations are what allow an imaginative world to become somewhere you can actually experience.

This is where Foundations begins.

A look beneath the surface

Foundations is a development series about the work that makes The CosmOS possible.

Each entry will take a closer look at one part of that work. We'll explore the ideas behind it, the problems encountered during development, the solutions being tested and what those decisions mean for the experience we're creating.

Some entries will examine working features. Others will follow experiments or unfinished work.

The distinction matters.

We want to share the process without presenting intentions as achievements, or promising that something is finished simply because we have a compelling vision for it.

Showing the work

The purpose of this series isn't to produce a catalogue of features.

It's to share how The CosmOS is actually being built.

That means showing real development, explaining decisions and being willing to discuss where things don't yet work.

It also means giving each subject the attention it deserves. An important system shouldn't be reduced to a screenshot and a sentence. It deserves an explanation of its purpose, its development and the experience it's intended to create.

Where we begin

There is a great deal to explore, and no need to cover it all at once.

We'll start beneath the surface, work through the foundations one subject at a time, and gradually reveal how those pieces contribute to the larger experience.

The CosmOS is a universe we intend to build, not simply describe.

Welcome to Foundations.

Feel before volume

First-wave testing is intended to gather useful feedback on focused, rough experiences. Readability, movement, interaction, and the desire to keep exploring matter more at this stage than a long list of promised content.

Registration is not immediate build access. It lets interested players be considered when a suitable test opportunity is ready; no specific launch date is announced here.

Useful feedback

• What is understandable without an extended tutorial?
• Which interactions feel clear, confusing, or inconsistent?
• Does a rough experience still invite curiosity?

Future test announcements should identify the actual build and entry conditions separately.