Can this be used for systems outside of energy and microgrids?

Yes. While it uses a microgrid as the physical substrate, the underlying dynamical patterns serve as a universal resolver for abstract software architecture and AI paradigm tensions.

Do I need prior Modelica experience to use the OMG Tension Resolver?

While knowledge of Modelica is beneficial, the skill provides standardized implementation patterns for interfacing with OpenModelica FMUs and the Microgrid Gym environment directly.

How does this skill improve AI optimization?

It implements SafeOpt, a gradient-free Bayesian optimizer that balances exploration (sampling) and exploitation (maximizing) while ensuring the system stays within defined safety boundaries during the process.

What is a 'tension' in the context of this Claude Code skill?

Tensions represent conflicting technical requirements in a system, such as the need for both discrete logic and continuous physics, or symbolic rules and neural learning. This skill provides a mathematical bridge to resolve them.

OMG Tension Resolver

Name: OMG Tension Resolver
Author: plurigrid

byplurigrid

•

Data Science & ML

Resolves architectural and algorithmic paradigm tensions using OpenModelica Microgrid Gym dynamics as a physical substrate.

The OMG Tension Resolver skill leverages the OpenModelica Microgrid Gym (OMG) to bridge conflicting technical paradigms—such as symbolic vs. subsymbolic reasoning or discrete vs. continuous control—by mapping them onto energy flow dynamics. By utilizing Functional Mock-up Units (FMUs), SafeOpt Bayesian optimization, and hierarchical control patterns, it provides a rigorous dynamical system framework for balancing opposing forces in complex software architectures. This tool is particularly useful for developers building hybrid AI systems that require formal stability guarantees alongside flexible, data-driven learning capabilities.

Key Features

01Resolves temporal vs. atemporal tensions via FMU simulation and steady-state Lyapunov analysis

02Coordinates local and global control through hierarchical multi-inverter coordination patterns

038 GitHub stars

04Balances exploration and exploitation using SafeOpt gradient-free Bayesian optimization

05Bridges discrete switching logic with continuous power flow modeling using PWM averaging techniques

06Integrates symbolic Modelica equations with subsymbolic neural network controllers via hybrid envelopes

Use Cases

01Optimizing complex systems where gradient-based backpropagation is unavailable or computationally expensive

02Designing hybrid AI controllers that require both neural flexibility and symbolic safety constraints

03Managing multi-agent coordination in systems with competing local constraints and global stability objectives

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add plurigrid/asi omg-tension-resolver

For use in Claude.ai and ChatGPT

Download Skill