langchain-coordinode 1.0.6

LangChain integration for CoordiNode — GraphStore backed by graph + vector + full-text

langchain-coordinode

LangChain integration for CoordiNode — GraphStore implementation and GraphCypherQAChain support for GraphRAG pipelines.

Installation

pip install langchain-coordinode

uv add langchain-coordinode

Requirements

• Python 3.11+
• Running CoordiNode instance

Quick Start

GraphCypherQAChain — Question Answering over a Knowledge Graph

from langchain_coordinode import CoordinodeGraph
from langchain.chains import GraphCypherQAChain
from langchain_openai import ChatOpenAI

# Connect to CoordiNode
graph = CoordinodeGraph("localhost:7080")

# Build a QA chain that generates and executes Cypher queries
chain = GraphCypherQAChain.from_llm(
    ChatOpenAI(model="gpt-4o-mini"),
    graph=graph,
    verbose=True,
)

result = chain.invoke({"query": "What concepts are related to attention mechanisms?"})
print(result["result"])

Schema Inspection

from langchain_coordinode import CoordinodeGraph

graph = CoordinodeGraph("localhost:7080")

# Refresh schema from database
graph.refresh_schema()

# Schema string used by the LLM to generate Cypher
print(graph.schema)
# Node properties: Person (name: String, age: Integer), Concept (name: String) ...
# Relationships: (Person)-[:KNOWS]->(Person), (Document)-[:ABOUT]->(Concept) ...

Direct Cypher Queries

from langchain_coordinode import CoordinodeGraph

graph = CoordinodeGraph("localhost:7080")

# Returns List[Dict[str, Any]]
result = graph.query(
    "MATCH (n:Person)-[:KNOWS]->(m) WHERE n.name = $name RETURN m.name AS colleague",
    params={"name": "Alice"},
)
for row in result:
    print(row["colleague"])

LLMGraphTransformer — Extract Knowledge from Text

from langchain_community.graphs.graph_document import GraphDocument
from langchain_openai import ChatOpenAI
from langchain_experimental.graph_transformers import LLMGraphTransformer
from langchain_coordinode import CoordinodeGraph
from langchain_core.documents import Document

graph = CoordinodeGraph("localhost:7080")
llm = ChatOpenAI(model="gpt-4o-mini", temperature=0)

transformer = LLMGraphTransformer(llm=llm)

docs = [Document(page_content="Alice knows Bob. Bob works at Acme Corp.")]
graph_docs = transformer.convert_to_graph_documents(docs)

# Store extracted entities and relationships
graph.add_graph_documents(graph_docs)

Connection Options

# host:port string
graph = CoordinodeGraph("localhost:7080")

# TLS
graph = CoordinodeGraph("db.example.com:7443", tls=True)

# Custom timeout
graph = CoordinodeGraph("localhost:7080", timeout=60.0)

API Reference

CoordinodeGraph

Method	Description
query(query, params)	Execute Cypher, returns List[Dict[str, Any]]
refresh_schema()	Reload node/relationship schema from database
add_graph_documents(docs)	Batch MERGE nodes + relationships from GraphDocument list
keyword_search(query, k, label, fuzzy, language)	Full-text BM25 search — returns [{"id", "score", "snippet"}, …]
similarity_search(query_vector, k, label, property)	Vector nearest-neighbour search — returns [{"id", "node", "distance"}, …]
schema	Schema string for LLM context
structured_schema	Structured schema dict for programmatic access

Related Packages

Package	Description
coordinode	Core gRPC client
llama-index-graph-stores-coordinode	LlamaIndex PropertyGraphStore

Links

• Source
• CoordiNode server
• LangChain docs
• Issues

Support

USDT (TRC-20): TFDsezHa1cBkoeZT5q2T49Wp66K8t2DmdA

GitHub Sponsors · Open Collective

License

Apache-2.0