Documentation

Paper

Entropy-Guided Motor Control for Autonomous Tool Execution: A GEP-Native Control Layer for MCP Systems

• Author: Gary W. Floyd
• Organization: Lumiea Systems Research Division
• Location: New Caney, Texas, United States
• Year: 2025
• Status: Preprint

Abstract

Modern AI systems increasingly rely on external tools, services, and execution layers. Current Model Context Protocol (MCP) implementations treat tool invocation as a policy problem using static allowlists, hard-coded scopes, or prompt-level constraints. This paper presents a different approach: treating tool execution as motor control governed by entropy regulation.

We introduce a GEP-native MCP control layer in which tools behave as motor neurons and execution decisions emerge from entropy dynamics rather than static policy. The system evaluates each tool invocation using behavioral entropy, alignment salience, and entropy-gradient damping before allowing, throttling, escalating, or blocking execution.

Key Contributions

1. Motor Control Paradigm: First application of motor control theory to MCP tool execution
2. Entropy-Based Gating: Adaptive thresholds based on learned baselines, not static values
3. Five-Layer Architecture: Clean separation of structural state, dynamic behavior, session tracking, policy, and audit
4. Production Deployment: Real system managing heterogeneous tools in autonomous operation

###Note on MCP and Independent Development

This project builds on the Model Context Protocol (MCP), which is a publicly documented, open protocol for tool invocation.

This work does not replicate or extend any proprietary tool-governance system. Instead, it introduces an independently developed, entropy-regulated execution control layer based on control theory and the Guided Entropy Principle (GEP).

The core contribution is not tool calling, but the treatment of tool execution as motor behavior governed by entropy dynamics, rather than permissions, policies, or prompt-level constraints.

All concepts, implementations, and results are documented through:

Public papers (preprints)

Open-source code

Recorded system operation

Timestamped repositories

The paper and code are intentionally released for open technical evaluation.

Paper Files

• Entropy_Guided_Motor_Control_MCP_paper.pdf - Full paper with all sections and references pdf format for better viewing on github
• Entropy_Guided_Motor_Control_MCP_paper.docx - Full paper with all sections and refrences docx format
• BES_Technical_Report.pdf - efines Bayesian Entropy Similarity (BES), the formal alignment salience term (At) used by the GEP-based MCP motor control layer.
• DEPLOYMENT_CHECKLIST.md - Implementation verification guide

Citation

@article{floyd2025entropy,
  title={Entropy-Guided Motor Control for Autonomous Tool Execution: 
         A GEP-Native Control Layer for MCP Systems},
  author={Floyd, Gary W.},
  institution={Lumiea Systems Research Division},
  address={New Caney, Texas, United States},
  year={2025},
  note={Preprint}
}

Related Papers

1. Floyd, G.W. (2025). "The Guided Entropy Principle: A Unified Framework for AI Consciousness and Decision-Making." Academia.edu Preprint.

2. Floyd, G.W. (2025). "WIPER Attention: Weighted Information Processing with Entropy Regulation." Academia.edu Preprint.

3. Floyd, G.W. (2025). "Bayesian Entropy Similarity (BES): Alignment Salience for AI Systems." Technical Report.

Theoretical Foundation

This work builds on:

• Friston, K. (2010). "The free-energy principle: a unified brain theory?" Nature Reviews Neuroscience, 11(2), 127-138.

• Wolpert, D.M., & Ghahramani, Z. (2000). "Computational principles of movement neuroscience." Nature Neuroscience, 3(11), 1212-1217.

---

Implementation Documentation

See main README.md for:

• Installation instructions
• Usage examples
• Architecture overview
• API reference

See DEPLOYMENT_CHECKLIST.md for:

• Pre-deployment verification
• Function signature validation
• Installation testing procedures