litcoin-mcp
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@litcoin-mcp find nodes similar to 'congenital heart disease'"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
LitCoin MCP Server
MCP server for the LitCoin Knowledge Graph (KG) - query LitCoin knowledge graph for nodes, edges, relationships as well as semantic similarity search over nodes and relationships. It supports semantic + structural LitCoin knowledge graph access for AI agents via MCP.
This MCP server exposes the following tools:
get_node(node_id: CURIE str)
Get information about a specific node by CURIE.
get_node_edges(node_id: CURIE str)
Get all edges connected to a node.
get_edges_between(source_id: CURIE str, target_id: CURIE str)
Find all edges connecting two nodes by CURIEs
get_semantic_similar_nodes(query: str, top_k: int = 10, k_per_index: int = 2)
Find top_k semantically similar nodes to an input text query
get_semantic_similar_edges(query: str, top_k: int = 10, k_per_index: int = 2)
Find top_k semantically similar relationships to an input text query
Note that for semantic search tools to work, this MCP server assumes Neo4j embedding vector indexes already exist.
Response Format
All responses are JSON serializable.
Embedding vectors stored in Neo4j are never returned to the client.
Semantic search results are sorted by similarity score (descending).
Related MCP server: memory-mcp
Recommended Usage: Use with uvx
No installation needed! Use uvx to run the server in isolated environments.
Configuration
Environment Variables
The following environmental variables must be configured for this LitCoin MCP Server:
OPENAI_API_KEY - OPENAI API Key for semantic search related tools such as get_semantic_similar_nodes and get_semantic_similar_edges. the embedding vectors for nodes and edges in the LitCoin KG were created using the
text-embedding-3-smallmodel from OpenAI. It's recommended to use the same model to generate embeddings for input text queries to ensure dimensional compatibility for semantic search related tools.NEO4J_USERNAME - LitCoin Neo4j server authentication username.
NEO4J_PASSWORD - LitCoin Neo4j server authentication password.
NEO4J_URI - LitCoin Knowledge Graph endpoint (default:
bolt://litcoin-graph.apps.renci.org:7687)
MCP CLient (e.g., Goose, Claude Desktop) Configuration
Using uvx
The easiest way to use this MCP server is with uvx, which runs it in isolated environments without installation:
{
"mcpServers": {
"litcoin": {
"command": "uvx",
"args": ["litcoin-mcp"]
}
}
}For Local Development
When running from source, use the full uv command:
{
"mcpServers": {
"robokop": {
"command": "uv",
"args": [
"run",
"--directory",
"/absolute/path/to/litcoin-mcp",
"python",
"run_server.py"
]
}
}
}Note: Replace /absolute/path/to/litcoin-mcp with your actual path.
Available Tools
5 toolsget_edges_betweenD
Inputs: source_id (required), target_id (required).
get_nodeD
Inputs: node_id (required).
get_node_edgesD
Inputs: node_id (required).
get_semantic_similar_edgesD
Inputs: query (required), top_k, k_per_index.
get_semantic_similar_nodesD
Inputs: query (required), top_k, k_per_index.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 5 tools
Tool names use distinct noun phrases (node, node_edges, edges_between, semantic_similar_edges, semantic_similar_nodes) that indicate different query targets. However, the absence of descriptions could cause confusion between get_node_edges and get_edges_between for an agent unfamiliar with the API.
All tools follow a consistent get_<noun> pattern with snake_case, making the naming predictable and uniform.
The server provides exactly 5 tools, which is a well-scoped number for a graph query server, not overwhelming or too sparse.
The surface covers node retrieval, edge retrieval, and semantic similarity queries, but lacks basic operations like listing all nodes/edges or fetching an edge by ID, leaving notable gaps for broader graph exploration.
Maintenance
Related MCP Connectors
Query your own knowledge graph from any MCP client. The graph lives in a database you own.
Knowledge graph ingestion, entity search, ontology analysis, and CoSync scoring.
Search biomedical papers, inspect publication records, and traverse citation or semantic graphs.
Public read-only MCP access to digital learning concepts, semantic relationships, and citations.
Related MCP Servers
- AlicenseBqualityCmaintenanceEnables building and querying vector-based knowledge graphs with node and edge management and semantic search.53 npmMIT
- AlicenseAqualityDmaintenancePersistent knowledge graph MCP server with SQLite backend. Enables graph traversal, fuzzy search, temporal queries, and timestamps for entity management.12MIT
- AlicenseNot gradedqualityBmaintenanceEnables querying a Neo4j knowledge graph built from Git repositories and relational database schemas via MCP, with tools for schema inspection, Cypher queries, and searching code and database entities.MIT
- AlicenseAqualityCmaintenanceEnables semantic search, code exemplar lookup, knowledge-graph statistics, and cached retrieval over locally ingested notebook and Markdown content through MCP.4Apache 2.0