Descrição
🔹 Full Description / Overview
LLM-Store is a central AI provider, routing and governance layer for Unreal Engine.
Instead of hard-coding OpenAI, Ollama, llama.cpp, Claude, Gemini or other providers directly into every editor tool, Blueprint or plugin, you define stable
AI task routes once and let LLM-Store decide which provider, model, fallback, policy, secret backend and cost rule should be used.
This makes AI integrations easier to manage across real projects: local models for sensitive data, cloud models for high-quality reasoning, mock routes for tests, cost tracking for budget control, and clear policies for development, CI and shipping builds.
LLM-Store is designed as infrastructure. It does not force a specific AI workflow on your project, but gives your own tools, Blueprints and plugins a clean C++ and Blueprint-accessible way to use AI safely and consistently.
🔹 Key Features
• Centralized Provider & Model Management
Create and manage AI providers and models from one editor UI.
Supported provider types include:
• Ollama
• llama.cpp
• LM Studio
• vLLM
• Hugging Face TGI
• OpenAI
• Anthropic / Claude
• Google Gemini
• Azure OpenAI
• Mistral
• OpenRouter
• Mock provider for tests and offline workflows
• Route-Based AI Access
Define stable task routes such as:
• docs.summarize
• asset.explain
• naming.review
• rag.embed
• agent.patch
Your tools call the route, not a specific model name. You can later switch providers, models or fallback behavior without rewriting every integration.
• Local, Cloud and Hybrid Workflows
Use local providers like Ollama, llama.cpp, LM Studio, vLLM or TGI for sensitive project data and offline work.
Use cloud providers like OpenAI, Claude, Gemini, Azure OpenAI, Mistral or OpenRouter where quality, context size or reasoning strength matter.
Combine both through route fallback chains and policies.
• Cost Tracking & Usage Analysis
The plugin records AI usage in a cost ledger and provides an editor Costs tab for analyzing:
• request count
• success count
• cache hits
• prompt tokens
• completion tokens
• estimated cost
• provider breakdown
• model breakdown
• route/task breakdown
• recent events
• CSV export
• Policies, Governance & CI Checks
Define per-route and per-build rules such as:
• cloud allowed / forbidden
• local-only routes
• maximum estimated cost
• maximum context size
• allowed or denied provider types
• shipping builds must use local or mock providers
Support
If you encounter issues or have feature requests, feel free to reach out via the Fab support section.
Feedback from tools programmers, technical artists, AI workflow builders and Unreal Engine teams is especially welcome. This plugin is designed to sit in real production pipelines where AI usage needs to be configurable, understandable and governable.
Discord Support Channel