Interface as Shared Taxonomy, Not Maintained Abstraction.
Part of the ai0S (Adaptive Independence) project family.
Human intent → [ skillcode ] → AI inspection & resolution → system-specific implementation
The human references a skillcode — a short semantic anchor such as
[skillcode]repo-initialdir. An AI that understands the actual environment
resolves the anchor into the appropriate implementation. The AI is the
translation layer; the skillcode is the agreement.
Not through abstraction, but through the capability of the executing instance. A fixed GUI, CLI wrapper, or API layer encodes assumptions about paths, flags, and environment structure that go stale. A skillcode encodes no such assumptions — the same anchor resolves correctly on any OS, in any directory layout, under any tooling generation.
Traditional interface
Human
↓
Fixed GUI / CLI / API
↓
Rigid abstraction
↓
System
ai0S-interface:
Human
↓
Intent & skillcodes
↓
AI understanding
↓
Adaptive implementation
↓
System
The same property carries over time: a new OS version, a new shell, or a different automation API does not change the skillcode. It also improves on its own — more capable models resolve the same unchanged anchors into better implementations.
| File | Content |
|---|---|
docs/CONCEPT.md |
The concept. Background, principles, and the two-stage evolution of the interaction model |
docs/FAQ.md |
Objections, clarifications, and dead ends — read before extending |
Idea/ |
Draft archive. The working notes this concept was distilled from; not normative |
| Project | Content |
|---|---|
| ai0S-adaptive-independence | Core concept and architectural foundations |
| ai0S-setup | AI-assisted installation of applications as reproducible modules — the first concrete implementation of the core concept |
Core concept. This repository defines principles only — concrete implementations live in separate repositories.
MIT. See LICENSE.
Engineered with Gemini. Authored with Meta Muse (contribution). Reviewed with Claude (feedback).