Intelligence Engine
Domain-agnostic intelligence engine with schema-driven knowledge graphs, hybrid search, and MCP server.
1261+ Tests · 15 MCP Tools · 33 REST Endpoints · 8 Languages · 2 Domains
Screenshots
![]() |
![]() |
| Knowledge Graph — 5000 nodes, 7048 edges | Cypher Console — 50 rows in 9ms |
![]() |
![]() |
| Dashboard Timeline — indexing history | Health Metrics — dead code, cycles, hubs |
![]() |
![]() |
| Cross-Project Compare — all indexed projects | AI Memory — KuzuDB + Minna browser |
Architecture
| Domain Schema | → | Extractors | → | Knowledge Graph | → | Hybrid Search | → | MCP / Web UI |
|---|---|---|---|---|---|---|---|---|
| YAML configs | Tree-sitter + Custom | KuzuDB per-domain | BM25 + Semantic + Graph | 15 tools · 33 endpoints |
Documentation
- Architecture Deep-Dive — Domain schema system, pipeline, storage modes
- Technical Overview — Parser, graph, search, AI, MCP, REST details
- Architectural Decisions — Key design choices and rationale
- Project Context — What it does, why it exists, status
Interactive Demos
- Which Search Strategy? — Choose the right search approach for your query
Highlights
- Schema-driven domains: define entity types, relationships, and search profiles in YAML
- Code intelligence: 8 languages parsed via Tree-sitter (Python, JS, TS/TSX, Java, Go, HTML, CSS)
- Archaeology MVP: first non-code domain validates domain-agnostic architecture
- Hybrid search: 3-way RRF fusion (BM25 + semantic + graph)
- Graph visualization: React + Sigma.js force-directed explorer
- AI integration: summaries and Q&A via Claude, OpenAI, Gemini, Ollama
- MCP server: 15 tools for AI coding assistants
- Multi-project: shared DB mode with cross-project Cypher queries





