| description | An essay-driven engineering archive on AI-assisted development, human judgment, and observable engineering work. |
|---|
Language: English | 한국어
Are you using AI?
Really?Do you trust what it produces?
Really?Have you inspected what it changed, what it broke, and what you are now responsible for?
Really?
AI can now produce code faster than humans can verify it.
Cosmic Horizon is a map for the work that remains human: judgment, verification, recovery, and responsibility.
In astronomy, the cosmic horizon is the outer boundary of the observable universe: the farthest limit from which signals can still reach us.
In this orbit, Cosmic Horizon names the entire observable world of AI-assisted development as I experience it—encompassing code, tools, prompts, documents, judgment, verification, recovery, and responsibility.
It is the boundary line where raw technical signals meet the human decisions that hold them together.
Cosmic Horizon is an essay-driven engineering archive for turning observations about software, AI-assisted work, and human judgment into coordinates that can be understood, refined, and returned to.
It focuses on structure, constraints, verification, recovery, and responsibility in software work shaped by AI.
This archive is for developers, technical leads, and architecture-minded practitioners who work with AI without giving up structure, judgment, or responsibility.
It is also my own landmark for returning to origin whenever the speed of my thinking outpaces observation.
A month of typhoons in the China Sea, a cracked mainmast, and a crew on half-rations.
Yet, it is all settled the moment the first chest is opened in London.
One sip of that black tea, and you realize the ocean was never our master.
This archive speaks in the language of navigation, orbit, signal, rations, and return.
These terms compress the raw, complex realities of engineering: uncertainty, constraints, drift, recovery, judgment, and responsibility. They serve as a precise vehicle to carry technical weight across different minds, roles, and depths of knowledge.
I enjoy reinterpreting my own understanding and reframing it for others. If these metaphors fail to land, it is simply because our focal points and perspectives diverge.
That, too, is by design.
What is your domain of choice? Bring it forward, and I will gladly re-map it into the metaphors of your own world.
This, too, is by design.
These pages may resemble essays at first glance, but they are more accurately understood as a map.
This archive does not need to be read from beginning to end like a book. It is a map to reopen when a coordinate needs to be checked again.
Some entries present perspective. Some define operating models. Others serve as field notes, templates, reusable records, and navigational instruments for the next round of execution.
Many entries in this archive are case-driven.
They begin with field friction, experiments, conversations, tools, or failures.
Those observations are compressed into language, then branch into operating models, protocols, cases, templates, and reusable records.
Rather than attaching examples to abstract principles, this archive turns lived cases into coordinates that can be returned to.
Documentation is not debris left after thought.
It is the act of turning the observation of an unfamiliar signal into a firm coordinate.
I write this archive so that insight does not disappear as a passing reaction.
When I observe something, I try to understand it.
When I understand it, I try to declare it in a form I can return to.
These pages are coordinates for returning to origin.
When judgment moves too fast, when language becomes vague, or when AI accelerates work faster than responsibility can follow, this archive becomes a landmark for returning to origin.
- Start with Ride, Don’t Race if you want the philosophy behind this archive.
- Read AI-Assisted Development Models if you want to understand the operating model.
- Read FTL-Bound Agents if you want to design AGENTS.md, CLAUDE.md, or similar agent instruction files as boundary systems for AI-assisted work.
- Explore Codex Chat Viewer if you want to see an earlier practical artifact from this orbit: a local-first viewer for Codex CLI session logs, built to inspect, review, document, and revisit AI-assisted work. A newer version continuing this direction has since been released as Codex JSONL Observatory, extending the work from transcript viewing into versionable worklog bundle export.
- Read Why My Ship Is Ivory to see how this archive turns an actual Figma MCP and Codex experiment into a record of execution, verification, and changing work boundaries.
- GitHub: @revertable
- LinkedIn: Riu Salze
- Published Site: Visit Cosmic Horizon
Cosmic Horizon was written and structured independently. Some surrounding themes may overlap with broader public discourse, but this archive does not claim ownership over general AI-assisted development discourse, third-party tools, public design materials, or concepts that may already exist elsewhere.
When quoting, summarizing, adapting, or referencing this archive’s original naming, metaphors, terminology, structure, boundary models, documented patterns, or blueprints, please provide visible attribution.
Preferred citation:
Riu Salze. (2026). Cosmic Horizon.
Published archive: https://riu-salze-studio.gitbook.io/cosmic-horizon/
Repository: https://github.com/revertable/cosmic-horizon
License: CC BY-NC-SA 4.0
For detailed license terms, see the section below.
This work is licensed under CC BY-NC-SA 4.0.
Cosmic Horizon is shared as a non-commercial, share-alike archive.
You may read, share, quote, translate, and adapt this archive for non-commercial purposes, provided that proper attribution is given, changes are indicated where applicable, and adapted material remains under the same license.
Commercial use, paid redistribution, paid training material, consulting material, productized reuse, or incorporation into commercial products or services requires prior written permission.