Plasma GraphRAG: Physics-Grounded Parameter Selection for Gyrokinetic Simulations

Plasma GraphRAG is a niche but highly specialized application of GraphRAG to the domain of fusion energy and plasma physics. Its defensibility (score 4) stems from the domain-specific ontology and the curated corpus of plasma physics literature required to make the system useful; however, the underlying technical framework is a standard implementation of GraphRAG patterns that are becoming commoditized. The unusual quantitative signal (0 stars but 9 forks in only 10 days) strongly suggests this is an academic research project with institutional backing (likely a specific plasma physics lab) where collaborators are actively working on the code even if it hasn't reached public 'viral' status. Frontier labs like OpenAI or Google are highly unlikely to compete here as the total addressable market is tiny—limited to specialized researchers at institutions like ITER, PPPL, or Max Planck. The primary risk is 'LLM Reasoning' evolution: as models gain better internal reasoning and larger context windows, the need for complex GraphRAG for parameter extraction may be displaced by simpler RAG or long-context prompting. The project’s moat is its physics-grounded knowledge graph, which is hard to build without subject matter experts (SMEs), but easy to replicate if those SMEs use more general-purpose tools like Microsoft's GraphRAG or WhyHow.ai.