Model Context Protocol
The Model Context Protocol (MCP) is a standard for connecting Large Language Models (LLMs) to platforms like Blue Morpho. This guide covers how to connect Blue Morpho to the following AI tools using MCP:
- Cursor
- Claude desktop
Once connected, your AI assistants can interact with and query your Blue Morpho projects on your behalf. To see a list of available tools, see here.
Step 1: Create a personal access token
First, go to your Blue Morpho settings and create an API key (personal access token). This will be used to authenticate the MCP server with your Blue Morpho account.
Step 2: Configure your AI tool
MCP-compatible tools can connect to Blue Morpho using the Blue Morpho remote MCP server.
Cursor
In Cursor settings, search for "mcp". In "Tools & Integrations", click on "New MCP Server".
This creates a mcp.json file:
{
"mcpServers": {
"blue-morpho": {
"type": "url",
"url": "https://app.getbluemorpho.com/mcp/",
"headers": {
"X-API-Key": "YOUR-API-KEY-HERE"
}
}
}
}
Replace YOUR-API-KEY-HERE with your Blue Morpho API key (or preferably, use environment variables).
Save the file. Navigate back to Settings → MCP. You should see a green active status once connected.
Claude desktop
Navigate to Settings → Developer → Edit Config:
{
"mcpServers": {
"blue-morpho": {
"command": "npx",
"args": [
"mcp-remote",
"https://app.getbluemorpho.com/mcp/",
"--header",
"X-API-Key: YOUR-API-KEY-HERE"
]
}
}
}
Replace YOUR-API-KEY-HERE with your Blue Morpho API key (or preferably, use environment variables).
Restart Claude. Navigate back to Settings → Developer → Edit Config. You should see your running MCP connection.
Step 3: Use Blue Morpho MCP tools
Once your AI tool is connected to Blue Morpho via MCP, you can immediately begin issuing commands to interact with your projects.
Try asking your AI assistant to list Blue Morpho projects to make sure everything is working properly!
Learn more about the Blue Morpho tools accessible through MCP below.
Project and Knowledge Base Management
list_projects
Lists all projects in the application.
Parameters:
None
Usage:
Useful for discovering available projects to work with.
list_knowledge_bases
Lists all knowledge bases within a specific project.
Parameters:
- project_id
Usage:
Helps you find the knowledge bases available for analysis.
Document Management
list_knowledge_base_documents
Lists the original documents used to build a knowledge base.
Parameters:
- project_id
- knowledge_base_id
Usage:
Shows you what source documents were processed to create the graph.
get_knowledge_base_document_urls
Provides download URLs for specific documents.
Parameters:
- project_id
- knowledge_base_id
- document_id
Returns:
Pre-signed AWS S3 URLs (expire after 30 minutes).
Usage:
Essential for accessing the original document content.
Graph Structure and Ontology
get_knowledge_base_ontology_version
Retrieves the ontology (schema) for a knowledge base.
Parameters:
- project_id
- knowledge_base_id
Usage:
Critical: Must be called before any graph querying to understand the data structure.
Graph Querying and Search
compute_cypher_query
Executes Cypher queries against the graph database.
Parameters:
- cypher_query
- project_id
- knowledge_base_id
Notes:
- Uses Memgraph’s Cypher implementation (openCypher standard).
- Prerequisite: Retrieve ontology first.
search_node_in_graph
Searches for nodes matching a query within the graph.
Parameters:
- search_query
- project_id
- knowledge_base_id
- node_type_name (optional, searches all types if not specified)
- max_number_results (optional, max 30)
Limitations:
- Cannot search SourceDocument nodes.
- Prerequisite: Retrieve ontology first.
Key Workflow
- Discovery:
- List projects → List knowledge bases
- Schema Understanding:
- Get ontology (mandatory before querying)
- Data Exploration:
- Search nodes or run Cypher queries
- Document Access:
- List documents → Get document URLs for original sources
Important Notes
- Ontology First: Always retrieve and display the ontology before performing any graph operations.
- Time Limits: Document URLs expire after 30 minutes.
- Query Limits: Node searches return a maximum of 30 results.
- Cypher Compatibility: Uses openCypher standard (no APOC or Neo4j-specific functions).
- Source Documents: Use the document URL tool instead of querying
SourceDocumentnodes directly.