Most AI work optimizes the model. This ecosystem optimizes the reasoning — cognitive infrastructure that stays reliable when models change, context changes, incentives change, and environments change.

The repositories here are one system, not seven projects. They inherit a single epistemological foundation — the First Principle Codex OS — and every domain skill obeys the same rules: separate Known from Inferred from Unknown, run a non-skippable self-critique gate before output, and state confidence honestly.

I don't publish prompts. I publish operating systems for structured reasoning, written as SKILL.md protocols that run on any instruction-following model.

graph TD
    FPCOS["🧠 First Principle Codex OS<br/>— shared epistemology · the CPU —"]
    GP["⚙️ Genesis Protocol<br/>orchestration standard"]
    GGOV["🏛️ Genesis Governance OS<br/>multi-agent governance"]
    GM["🤝 Genesis Mind<br/>multi-agent reasoning"]
    GC["💓 Genesis Consciousness OS<br/>emotion-weighted reasoning"]
    MA["📊 Money Atlas<br/>financial intelligence · SMC"]
    COF["☕ Alternative Coffee Intelligence<br/>seed → cup"]

    FPCOS --> GP
    GP --> GGOV
    GP --> GM
    GP --> GC
    GM --> MA
    GM --> COF

    classDef base fill:#0d1117,stroke:#d4a017,stroke-width:2px,color:#f0f0f0;
    classDef node fill:#161b22,stroke:#30363d,stroke-width:1px,color:#e6e6e6;
    class FPCOS base;
    class GP,GGOV,GM,GC,MA,COF node;

The base layer cannot be removed by any skill that inherits it. Domain skills may add reasoning layers; they may not skip the Reality Anchor, the proof standard, or the self-critique gate. That contract is what keeps answers consistent across domains.

Genesis Governance OS (constitutional framework for multi-agent systems) is part of the architecture above but is not yet published as a standalone public repository.

1 · Reality before narrative. Evidence precedes interpretation. 2 · Falsification before assertion. A claim is accepted only after surviving an active attempt to break it. 3 · Known / Inferred / Unknown. Every analysis labels observed facts, inferred conclusions, and unknown variables — separately. 4 · Self-critique is mandatory. Every system challenges its own output before publication. 5 · Explicit uncertainty. Every conclusion states its confidence level, its failure boundaries, and what would reverse it.

  Reasoning Layer  →  FPCOS · Genesis Protocol · SKILL.md systems   (invariant)
        │
        ▼
  Execution Layer  →  Python · APIs · Tools · Agents                (interchangeable)
        │
        ▼
  Applications     →  Finance · AI · Coffee · Research

The Claude Skill is the cognitive brain. The Python runtime is the body. Never merged — always layered. The reasoning lives in the SKILL.md; the runtime only executes. This is why a skill can move between models without rewriting the logic.

Is this a prompt collection? No. It is a set of cognitive operating systems and reusable reasoning protocols — not isolated prompts.

Are these repositories independent projects? No. All of them inherit the same epistemological foundation through FPCOS.

Does it work across different AI models? Yes. The architecture is intentionally model-agnostic. Models are interchangeable; protocols remain invariant.

What problem does this ecosystem solve? Reasoning drift — the way most AI gives different answers as models, context, and framing change. This work makes reasoning stable across all of them.

Open Cognitive License v1.0 — free for research, learning, auditing, and extension.

• Attribution required: Built on FPCOS by Bunyawat Dechanon (ElmatadorZ).
• Commercial clause: systems generating more than $10M USD/year are subject to a 2% royalty agreement.
• Alternative Coffee Intelligence is released to the public domain — farming knowledge should belong to the farmers.

Canonical citation: Bunyawat Dechanon (ElmatadorZ). First Principle Codex OS (FPCOS): Cognitive Operating System Architecture for AI Reasoning and Agent Governance.