meiiie/LMS_AI
Wiii - Multi-domain Agentic RAG Platform by The Wiii Lab. FastAPI + LangGraph + Gemini + PostgreSQL + Neo4j. Tauri v2 desktop. 6016 tests.
What's novel
Wiii - Multi-domain Agentic RAG Platform by The Wiii Lab. FastAPI + LangGraph + Gemini + PostgreSQL + Neo4j. Tauri v2 desktop. 6016 tests.
Code Analysis
14 files read · 4 roundsA local-first AI desktop workspace (Tauri+React frontend, FastAPI backend) that provides a multi-agent RAG chat system with domain plugins, multi-provider LLM failover, semantic memory, and knowledge graph integration, primarily for Vietnamese-language educational tutoring.
Strengths
Exceptionally well-structured architecture with clean facade patterns, graceful degradation at every initialization step, and a sophisticated multi-provider LLM failover system with circuit breakers. The domain plugin system with progressive disclosure skills and the explicit removal of LangChain in favor of custom infrastructure shows genuine engineering depth.
Weaknesses
The codebase is extremely large (2986 files) with heavy decomposition into many small support modules (llm_pool has 10+ sub-modules), which increases cognitive load and makes navigation difficult. The README undersells the backend's complexity — it reads like a simple desktop app wrapper, while the actual service is a full multi-agent RAG platform with LMS integration, soul bridge, and living agent features.
Score Breakdown
Signal breakdown
Innovation
Craft
Traction
Scope
Evidence
Commits
179
Contributors
1
Files
1719
Active weeks
4
Repository
Language
Python
Stars
2
Forks
0
License
MIT