ARC OS is a 5-layer logic framework designed to create a shared language of thought between humans and AI. It's model-free (no weights or prompts required), domain-agnostic, and fully auditable—ideal for building traceable decisions, reducing bias, and bridging symbolic reasoning with LLMs. Think of it as infrastructure for transparent AI workflows.
This repo contains the core specs, examples, and snapshots. It's early-stage, so it may not be user-friendly yet, but you can test it manually by pasting files into any LLM (e.g., ChatGPT, Claude, Grok, Gemini).
- Layer 1: Input Normalization (Muay Glasses): Normalizes data into numbers (adaptable beyond Muay Thai to any domain).
- Layer 2: Prediction Structure (Seannoi Core): Calculates logic tilt % (not probabilistic prediction—focus on balanced reasoning).
- Layer 3: Meta-Intent Oversight (Advisor Layer): Audits Layers 1 and 2 for consistency and intent.
- Logic Renderer: ARC Builder: Generates structured logic trees, self-checks, and outputs that both AI and humans can understand and audit.
- Meta-Layer Audit Builder: ARC Supervisor: Ensures overall framework integrity.
- Snapshots: Real-use simulations (e.g., ElonGov, Grok) showing field remapping for cross-domain applications—you can use ARC Builder to remap fields or integrate Layers 1/2.
Each layer has a unique role, making the stack modular. It's logic-based for transparency, works with most AI models/agents, and can be deployed (with permission).
-
Main Demo:
-
Additional Short Demo:
- Download
arc_logic_stack_v1.5.zipfrom the releases or clone this repo. - Paste or attach a spec file (e.g.,
ARC_Builder_EN_v1.0_LICENSED.md) into an LLM. - Prompt: "Use this as the reasoning framework" or "Be this one" for testing.
- Ask queries like "Render in L3" or any logic-based question—the LLM will output structured responses (logic trees, self-checks, etc.).
- Results vary by LLM—it's manual for now, but powerful for audits.
If it helps with your prompts or decisions, let me know!
- Do I need coding skills? No—just paste into an LLM for testing.
- Is it a full app? No, specs for manual testing or building tools.
- Different from GPT? Yes—adds auditable structure before AI responds.
- Works on web? No, but you can build it into one (permission required).
Each file is licensed under LICENSE.md (Logic Attribution v1.5 – NC / ND / DeployRestricted).
- Attribution required.
- No commercial use or deployment without written permission.
- Forks for private use only; no publication or AI integration.
- Contact: [email protected] or @autononthagorn on X.
| License Tier | Includes | Price |
|---|---|---|
| Evaluation License | Read-only, internal/sandbox use | $0 |
| Standard Deploy | Use in agents or closed systems | $1,500 |
| Institutional License | For research/ethics/critical tools | $4,000 |
| Exclusive License | Full control, modifiable, exports | Contact |
Paid licenses unlock building/deploying with ARC OS specs (e.g., GPT agents, decision tools). Evaluation is free for manual testing.
- Email: [email protected]
- X (Twitter): @autononthagorn
⭐️ If you find it useful, star the repo!
📧 Cloned? Email or DM feedback—your input shapes the next version. (e.g., "Tried the .md specs? What worked/missed?")
"I finally understood what GPT missed before." — Early user feedback
Logic Attribution specs & examples for ARC OS. Built by a solo dev in 2 weeks—no prior coding background. Early stage, so feedback welcome! ❤️