GraphKnowledge

graph-studio

Common MCP Workflows

The following examples represent some common MCP workflows.

Loading Data Files

Example prompt: "I have three JSON files in the root of the default file connection: customers.json, orders.json, and products.json. Create a graphmart called 'E-Commerce Data' that loads these files, then activate it."

Result: The assistant discovers the graphmart skill, resolves the create and activate tools, and orchestrates the workflow.

Querying Data

Once a graphmart is active:

Example prompts: "How many customers are in the E-Commerce Data graphmart?"; "Show me the top 10 products by order count."

Result: The assistant resolves the query tools and translates your questions into SPARQL.

Creating Ontologies

Example prompt: "Look at the data in my E-Commerce graphmart and create an ontology that models customers, orders, and products."

Result: The assistant queries the data, designs an ontology with appropriate classes and properties, saves it to Graph Studio, and registers it.

Transforming Data

Example prompt: "Create a transformation layer in my E-Commerce graphmart that maps the raw JSON data into the ontology we just created."

Result: The assistant creates a new layer with mapping templates connecting source properties to the target ontology.

Discovering Relationships

Example prompt: "Find links between all the datasets in my E-Commerce graphmart."

Result: The assistant analyzes property values across datasets to find overlapping data and creates link steps.

Capturing Workflow as a Skill

After working through a multi-step problem, you can ask the assistant to save the approach as a reusable skill.

Step Example Prompt
Solve the problem "Load the CSV files from the sales folder, profile the data, generate an ontology, and build a graphmart."
Review what was done "Summarize the steps you just took."
Create the skill "Create a skill called 'Sales Data Onboarding' that captures this workflow."
Refine "Add a sub-skill for the data profiling step so I can reuse that part separately."
Reuse later "Run the Sales Data Onboarding skill on the Q2 data folder."

The assistant will create the skill with an appropriate description, link the relevant tools it used, and optionally break it into sub-skills for the distinct phases. The new skill appears immediately in the server's Skills tab under Agentic Skills.

Source: https://docs.sw.siemens.com/documentation/external/PL20260212925461721/en-US/graph_studio/mcp-common-workflows1.htm · retrieved 2026-08-23