Skip to content
LOPS Docs

Graphiti Retriever Operator

The Graphiti Retriever operator provides powerful query capabilities for Graphiti knowledge graphs. It supports multiple search strategies including semantic search, keyword search, and hybrid approaches to find relevant facts, entities, and relationships from your knowledge graph.

Graphiti Retriever enables:

  • Multi-Strategy Search: Semantic, keyword, graph traversal, and hybrid search methods
  • Entity & Relationship Retrieval: Find both individual entities and their relationships
  • Namespace Filtering: Search within specific project or user namespaces
  • Agent Integration: Exposed as a tool for automated knowledge graph queries
  • Flexible Ranking: Configurable result limits and relevance scoring

Agent Tool Integration

Section titled “Agent Tool Integration”
🔧 GetTool Enabled 1 tool

This operator exposes 1 tool that allow Agent and Gemini Live LOPs to search knowledge graphs using semantic and keyword approaches to retrieve relevant information for AI workflows.

Use the Tool Debugger operator to inspect exact tool definitions, schemas, and parameters.

The Graphiti Retriever automatically exposes a search_knowledge_graph tool when connected to an Agent, enabling intelligent knowledge graph queries and retrieval.

Key Features

Section titled “Key Features”

Search Strategies

Section titled “Search Strategies”
  • Semantic Search: Uses embeddings for conceptual similarity matching
  • Keyword Search: BM25 algorithm for exact term and phrase matching
  • Hybrid Search: Combines semantic and keyword approaches for comprehensive results
  • Graph Traversal: Explores entity relationships and connections (planned)

Result Types

Section titled “Result Types”
  • Edges: Relationships between entities (facts and connections)
  • Nodes: Individual entities with their properties and summaries
  • Ranked Results: Relevance-scored results with detailed metadata
  • Namespace Filtered: Results scoped to specific knowledge spaces

Parameters

Section titled “Parameters”

Retriever Configuration

Section titled “Retriever Configuration”
Indexsource (Indexsource) op('graphiti_retriever').par.Indexsource OP
Default:
None
Connect (Connect) op('graphiti_retriever').par.Connect Pulse
Default:
None
Connectionstatus (Connectionstatus) op('graphiti_retriever').par.Connectionstatus Toggle
Default:
None
Queryphrase (Queryphrase) op('graphiti_retriever').par.Queryphrase String
Default:
None
Searchmethod (Searchmethod) op('graphiti_retriever').par.Searchmethod Menu
Default:
Semantic
Graphhops (Graphhops) op('graphiti_retriever').par.Graphhops Integer
Default:
2
Topk (Topk) op('graphiti_retriever').par.Topk Integer
Default:
5
Query (Query) op('graphiti_retriever').par.Query Pulse
Default:
None
Status (Status) op('graphiti_retriever').par.Status String
Default:
None

Namespace Management

Section titled “Namespace Management”
Usenamespace (Usenamespace) op('graphiti_retriever').par.Usenamespace Toggle
Default:
true

Usage

Section titled “Usage”

Basic Setup

Section titled “Basic Setup”
  1. Connect to Graphiti Index: Set Index Source parameter to reference your Graphiti Index operator
  2. Configure Search: Set search method and result limits
  3. Connect: Use Connect pulse to establish the connection

Search Operations

Section titled “Search Operations”
  1. Enter Query: Type search query in ‘Query Phrase’ parameter
  2. Select Method: Choose Semantic, Keyword, or Hybrid (recommended)
  3. Set Limit: Adjust ‘Top K’ parameter (default: 10)
  4. Execute: Click ‘Query’ button - results appear in ‘results_table’ output DAT
🔧 GetTool Enabled 1 tool

This operator exposes 1 tool that allow Agent and Gemini Live LOPs to interact with this operator.

Use the Tool Debugger operator to inspect exact tool definitions, schemas, and parameters.

The operator automatically exposes a search_knowledge_graph tool when connected to an Agent.

Search Methods

Section titled “Search Methods”

Semantic Search: Embedding-based similarity for conceptual queries

  • Example: “renewable energy solutions” → finds solar, wind, hydroelectric content

Keyword Search: BM25 algorithm for exact term matching

  • Example: “PyTorch” → finds exact mentions of PyTorch

Hybrid Search: Combines semantic and keyword (recommended)

  • Example: “machine learning Python” → finds both conceptual ML and Python-specific content

Output

Section titled “Output”

Results Table columns:

  • query: The search query executed
  • type: ‘edge’ (relationship) or ‘node’ (entity)
  • rank: Result ranking (1 = most relevant)
  • score: Relevance score
  • details: JSON object with complete result metadata

Result Types:

  • Edge Results: Relationships between entities (e.g., “causes”, “relates to”)
  • Node Results: Individual entities (e.g., “Solar Energy”, “Climate Change”)

Configuration Options

Section titled “Configuration Options”

Namespace Filtering:

  • Project-Scoped: Use Namespace ON - searches within current project
  • Global: Use Namespace OFF - searches across all namespaces

Query Optimization: Use Top K parameter to balance comprehensiveness vs speed

Integration

Section titled “Integration”

With Agent Operator

Section titled “With Agent Operator”
  1. Wire to Agent’s tool input - automatically detects search_knowledge_graph tool
  2. Add search instructions to Agent’s system prompt
  3. Agent will automatically search knowledge graph when relevant

With Chat Applications

Section titled “With Chat Applications”
  • Wire between user input and chat response
  • Use results to augment chat context with relevant facts and entities

Troubleshooting

Section titled “Troubleshooting”

No Results Found: Try different search methods, check namespace filtering, increase Top K value

Connection Problems: Verify Graphiti Index is configured and connected

Performance Issues: Reduce Top K value, use namespace filtering, check Neo4j indices

Section titled “Related Operators”

Research & Licensing

Zep (getzep)

Zep (getzep) is a technology company focused on building memory and knowledge management systems for AI agents. They specialize in creating practical, scalable solutions for agent memory persistence and contextual understanding in conversational AI applications.

Graphiti: Temporal Knowledge Graph Architecture

Graphiti is a temporal knowledge graph framework designed for AI agent memory systems. It automatically extracts entities and relationships from conversations and documents, creating a dynamic knowledge graph that evolves over time. The framework enables sophisticated retrieval and reasoning capabilities for enhanced agent performance.

Technical Details

  • Neo4j Backend: Leverages graph database for efficient relationship storage and querying
  • Temporal Modeling: Captures time-based relationships and entity evolution
  • Embedding Integration: Combines graph structure with vector embeddings for hybrid search

Research Impact

  • Agent Memory Systems: Advancing persistent memory capabilities for AI agents
  • Knowledge Graph Applications: Practical framework for temporal knowledge representation
  • Open Source Innovation: Democratizing access to advanced knowledge graph technologies

Citation

@misc{cui2025zep,
  title={Zep: A Temporal Knowledge Graph Architecture for Agent Memory},
  author={Cui, Cheng and Sun, Ting and Lin, Manhui and Gao, Tingquan and Zhang, Yubo and Liu, Jiaxuan and Wang, Xueqing and Zhang, Zelun and Zhou, Changda and Liu, Hongen and Zhang, Yue and Lv, Wenyu and Huang, Kui and Zhang, Yichao and Zhang, Jing and Zhang, Jun and Liu, Yi and Yu, Dianhai and Ma, Yanjun},
  journal={arXiv preprint arXiv:2501.13956},
  year={2025},
  url={https://arxiv.org/abs/2501.13956}
}

Key Research Contributions

  • Temporal knowledge graph architecture for dynamic agent memory
  • Automated entity and relationship extraction from unstructured text
  • Graph-based retrieval system for enhanced contextual understanding

License

Apache 2.0 License - This model is freely available for research and commercial use.