Aegntic MCP Servers

An elite Retrieval-Augmented Generation (RAG) system that transforms Obsidian vaults into AI-paired cognitive workflow engines with advanced Graphiti knowledge graph integration.

- L1: Semantic Context(30% weight) - Vector similarity search with OpenAI embeddings
- L2: Knowledge Graph(25% weight) - Graphiti-powered entity and relationship retrieval
- L3: Graph Traversal(15% weight) - NetworkX-based link traversal
- L4: Temporal Context(15% weight) - Time-based relevance and freshness
- L5: Domain Specialization(15% weight) - Context-aware retrieval
- L6: Meta-Knowledge(remaining weight) - Knowledge about knowledge

- 27+ Entity Types: concepts, people, organizations, technologies, methodologies, frameworks, algorithms, etc.
- 40+ Relationship Types: implements, uses, depends_on, extends, based_on, similar_to, integrates_with, etc.
- Dual-Graph Architecture: Neo4j (structured) + NetworkX (unstructured backup)
- Automatic Entity Extraction: Pattern matching and NLP-based entity recognition
- Relationship Detection: Confidence scoring and validation

- Claude Code Compatible: Full Model Context Protocol server implementation
- Tool-based API: Ingest, query, search knowledge graph, get entity context
- Real-time Status: System health monitoring and database connection checks
- Async Processing: High-performance concurrent operations

- Python 3.9+
- Docker & Docker Compose
- OpenAI API key
- Obsidian vault(optional but recommended)
- Neo4j Database(handled by setup scripts)
- Qdrant Vector Database(handled by setup scripts)

Option 1: Install from PyPI (Recommended)

git clone https://github.com/aegntic/aegntic-MCP.git cd aegntic-MCP/obsidian-elite-rag pip install -e .

# Initialize the system obsidian-elite-rag-cli setup # Start both databases (Qdrant + Neo4j) obsidian-elite-rag-cli start-databases # Or start manually with Docker docker run -d --name qdrant -p 6333:6333 -v $(pwd)/data/qdrant:/qdrant/storage qdrant/qdrant:latest docker run -d --name neo4j -p 7474:7474 -p 7687:7687 -v $(pwd)/data/neo4j:/data \ --env NEO4J_AUTH=neo4j/password --env NEO4J_PLUGINS='["apoc","graph-data-science"]' \ neo4j:5.14

# Ingest all markdown files obsidian-elite-rag-cli ingest /path/to/your/obsidian/vault # Check system status obsidian-elite-rag-cli status /path/to/your/obsidian/vault

# Start the MCP server for Claude Code integration obsidian-elite-rag-cli server

Add to your Claude Code configuration (~/.config/claude-code/config.json):

{ "mcpServers": { "obsidian-elite-rag": { "command": "obsidian-elite-rag-cli", "args": ["server"], "env": { "OPENAI_API_KEY": "your-openai-api-key" } } } }

# Query the RAG system obsidian-elite-rag-cli query "How does the RAG system work?" /path/to/vault # Search knowledge graph for entities obsidian-elite-rag-cli graph /path/to/vault --entity-query "machine learning" # Technical queries obsidian-elite-rag-cli query "JWT authentication patterns" /path/to/vault --query-type technical # Research queries obsidian-elite-rag-cli query "latest developments in LLMs" /path/to/vault --query-type research

When connected to Claude Code, you'll have access to these tools:
- ingest_vault- Ingest markdown files from an Obsidian vault
- query_rag- Query the elite RAG system with multi-layer retrieval
- search_knowledge_graph- Search the Graphiti knowledge graph for entities
- get_entity_context- Get rich context for a specific entity
- get_related_entities- Get entities related through relationships
- get_system_status- Get system status and database connections

@obsidian-elite-rag please ingest my vault at /Users/me/Documents/Obsidian @obsidian-elite-rag query "what are the key concepts in machine learning?" with vault path /Users/me/Documents/Obsidian @obsidian-elite-rag search_knowledge_graph for "neural networks" in vault /Users/me/Documents/Obsidian

┌─────────────────┐ ┌─────────────────┐ ┌─────────────────┐ │ Obsidian │ │ Claude Code │ │ MCP Protocol │ │ Vault │◄──►│ Integration │◄──►│ Server │ └─────────────────┘ └─────────────────┘ └─────────────────┘ │ ▼ ┌─────────────────────────────────────────────────────────────────┐ │ Elite RAG System │ ├─────────────────┬─────────────────┬─────────────────────────────┤ │ Semantic │ Knowledge │ Temporal & Domain │ │ Search │ Graph │ Specialization │ │ (Qdrant) │ (Neo4j) │ │ └─────────────────┴─────────────────┴─────────────────────────────┘

- Core: concept, person, organization, event, location
- Technical: technology, algorithm, framework, system, application
- Process: methodology, workflow, process, pattern
- Implementation: tool, library, database, api, protocol
- Documentation: standard, specification, principle, theory, model
- Architecture: design, implementation, project, research

- Structural: part_of, implements, extends, based_on, depends_on
- Semantic: similar_to, contrasts_with, related_to, examples_of
- Functional: uses, enables, requires, supports, improves
- Cognitive: defines, describes, explains, demonstrates, teaches
- Development: builds_on, applies_to, references, cites, tests
- Operational: manages, monitors, deploys, configures, maintains

- Retrieval Speed: <100ms for context-rich queries
- Knowledge Coverage: 95%+ recall on domain-specific queries
- Entity Recognition: 90%+ accuracy for concepts, people, organizations
- Relationship Extraction: 85%+ accuracy for semantic relationships
- Graph Traversal: <50ms for entity relationship queries up to depth 4
- Automation Coverage: 80%+ routine knowledge tasks automated

# Required OPENAI_API_KEY=your-openai-api-key # Optional (auto-configured by setup scripts) NEO4J_URI=bolt://localhost:7687 NEO4J_USER=neo4j NEO4J_PASSWORD=password QDRANT_HOST=localhost QDRANT_PORT=6333

The system usesconfig/automation-config.yamlfor detailed configuration:

knowledge_graph: enabled: true provider: graphiti graphiti: neo4j_uri: bolt://localhost:7687 neo4j_user: neo4j neo4j_password: "password" rag_system: layers: semantic: weight: 0.3 similarity_threshold: 0.7 knowledge_graph: weight: 0.25 max_depth: 4 # ... other layers

The system works best with this Obsidian vault structure:

00-Core/ # 🧠 Foundational knowledge 01-Projects/ # 🚀 Active work 02-Research/ # 🔬 Learning areas 03-Workflows/ # ⚙️ Reusable processes 04-AI-Paired/ # 🤖 Claude interactions 05-Resources/ # 📚 External references 06-Meta/ # 📊 System knowledge 07-Archive/ # 📦 Historical data 08-Templates/ # 📋 Note structures 09-Links/ # 🔗 External connections

We welcome contributions! Please see ourContributing Guidefor details.

# Clone the repository git clone https://github.com/aegntic/aegntic-MCP.git cd aegntic-MCP/obsidian-elite-rag # Install in development mode pip install -e ".[dev]" # Run tests pytest # Run with coverage pytest --cov=obsidian_elite_rag # Code formatting black src/ mypy src/

This project is licensed under the MIT License - see theLICENSEfile for details.

Created by:Mattae CooperEmail:research@aegntic.aiOrganization:Aegntic AI (https://aegntic.ai)

- Documentation:Project Wiki
- Issues:GitHub Issues
- Discussions:GitHub Discussions
- Email:research@aegntic.ai

- Graphiti- Knowledge graph construction for LLMs
- Qdrant- Vector similarity search engine
- Neo4j- Graph database
- LangChain- LLM application framework
- Model Context Protocol- Standard for AI tool integration

Hosted MCP server and coordination layer for AI coding agents — live API contracts, database schema, frontend/backend mismatch detection, and shared handoff tickets for Claude Code, Cursor, Codex, and Lovable.