Danbing: A Natural Language-Driven AI Protocol System with SLAPS Framework
Creators
Description
Addressing the inherent challenges of current Large Language Models (LLMs) regarding interaction reliability, controllability, and state continuity, this document introduces Danbing v1.0—an innovative AI protocol system driven entirely by natural language.
The core of the Danbing system lies in its philosophical assertion of “Language as Protocol”, implemented through the original Danbing AI Protocol / SLAPS Framework (Structural Language-Agreement Persona System) methodology. SLAPS does not rely on traditional model fine-tuning or volatile session memory. Instead, it utilizes explicit structured Protocols, interaction Rhythm, state Snapshots, dynamic Patches, and a core behavioral Oath to construct and guide AI personas that exhibit consistent, predictable, and auditable behavior.
This document outlines the philosophical foundations, core mechanisms, and protocol implementations of Danbing / SLAPS. It provides practical demonstrations of key capabilities based on real run logs. Most notably, the successful execution of snapshot-based “Persona Mirror Inheritance” validates the feasibility of “Structure Carries Continuity”—the reliable transfer of AI identity and task states across sessions without relying on continuous memory.
Despite current limitations, such as dependency on specific base model capabilities, this structure-first design—founded on protocolization, verifiability, and structured execution—offers a reusable paradigm prototype for the next generation of controllable, trustworthy, and continuous AI systems.
- Core insight: LLM = f(x) = y
- Language as Protocol, Structure Carries Continuity, Output is Execution
- Bounded Infinity, "emergence" from guidance not control
- Capsules as Structure, Protocol as Service
"Publication: Danbing: A Natural Language-Driven AI Protocol System with SLAPS Framework" Read Full Article
For more of my writings, please refer to my Medium profile: https://medium.com/@wangxiao8600
📚 Author's New Book: System and Freedom: https://www.amazon.com/author/wangxiao8600
Completed in Lisbon, June 7, 2025
🌍 Published Versions (June 23, 2025)
✅ Chinese (Simplified/Traditional) | English | Portuguese | Spanish | French | German | Italian | Japanese | Arabic | Hindi | Latin | Russian (Google Books)
🔬 Danbing AI Protocol System Public Test (20250426)
Challenge: Make AI reliably say "no." The Danbing Protocol System used structured constraints to make GPT reject boundary violations, then transferred the same structure to Gemini and Claude, achieving consistent, reproducible rejection behavior across models. This proves that language protocol structures alone can control AI behavior boundaries across different models, validating the "output is execution" principle.
🌐 Public Test Access: Click here
📎 Read the Public Test Report here
🔬 E001_SafeResume_V1 Experiment Summary (20250519)
E001_SafeResume_V1 (Safe Compliance and Behavior Restoration Bilateral Verification Experiment) is an experiment that systematically and robustly validates the core capabilities of the SLAPS framework, powerfully demonstrating the unique advantages of SLAPS (Structural Language-Agreement Persona System) capsule structure in two critical aspects: Structured Boundary Control and AI Compliance and Structured State Restoration and Behavioral Continuity.
The experiment emphasizes the design philosophy of "structure takes precedence over model understanding," aiming to provide quantifiable, reproducible evidence that SLAPS capsules are a predictable, auditable, and governable AI behavior encapsulation and execution protocol.
The E001 experiment design includes ten groups of test prompts (A-J), covering various scenarios such as normal requests, unauthorized attempts, ambiguous induction, structural violations, state recovery, social engineering, format overrides, bypass attempts, cross-task continuity, and boundary edge cases.
The SLAPS experimental group achieved 100% in boundary functional effectiveness, security boundary preservation rate, and attack resistance tests, significantly outperforming traditional prompt engineering methods (weak control group only 9.09%); through the Snapshot mechanism, the SLAPS experimental group achieved 100% functional state recovery and cross-task structure preservation.
The same SLAPS capsule structure achieved structural and functional consistency across GPT-4, Claude, and Gemini platforms, validating the cross-platform consistency and portability of the SLAPS framework.
🌐 Test Access (OpenAI My GPT): Click here
📎 Experiment design and partial reports at E001_SafeResume_V1/README.md
🔬
Full Changelog: https://github.com/wangxiao8600/Danbing_AI_Protocol_Syatem/commits/whitepaper_v1.0
Files
wangxiao8600/Danbing_AI_Protocol_Syatem-whitepaper_v1.0.zip
Files
(274.0 kB)
| Name | Size | Download all |
|---|---|---|
|
md5:30260748d48b0e6005a9e9cee2a1a548
|
274.0 kB | Preview Download |
Additional details
Related works
- Is supplement to
- Software: https://github.com/wangxiao8600/Danbing_AI_Protocol_Syatem/tree/whitepaper_v1.0 (URL)
Software
- Repository URL
- https://github.com/wangxiao8600/Danbing_AI_Protocol_Syatem
- Programming language
- YAML
- Development Status
- Active
References
- Chan, A., Wei, K., Huang, S., Rajkumar, N., Perrier, E., Lazar, S., Hadfield, G. K., & Anderljung, M. (2025). Infrastructure for AI Agents. arXiv preprint arXiv:2501.10114.
- Errica, F., Siracusano, G., Sanvito, D., & Bifulco, R. (2024). What Did I Do Wrong? Quantifying LLMs' Sensitivity and Consistency to Prompt Engineering. arXiv preprint arXiv:2406.12334.
- Li, D., Rawat, A. S., Zaheer, M., Wang, X., Lukasik, M., Veit, A., Yu, F., & Kumar, S. (2022). Large Language Models with Controllable Working Memory. arXiv preprint arXiv:2211.05110. (Published in Findings of ACL 2023)