Powered by Zoomin Software. For more details please contactZoomin

Semaphore 5.13 MCP Server Guide

Using MCP Tools: Examples

  • Last Updated: November 20, 2026
  • 5 minute read
    • Semaphore
    • Version 5.13
    • Documentation

After connecting an MCP-compatible AI client to the Semaphore MCP Server, you can use natural-language prompts to explore and manage Knowledge Model Management (KMM) resources.

Instead of selecting individual MCP tools manually, you can describe what you want to accomplish. Your AI client can then select and invoke the appropriate tools exposed by Semaphore.

This guide provides examples of common tasks, from finding a knowledge model to retrieving concept information and preparing model changes.

Before you begin

Make sure you have:

  • Access to a Semaphore environment with the MCP Server available.
  • A configured MCP client connected to Semaphore.
  • A valid Semaphore access token.
  • Permission to access the knowledge models you want to work with.

For setup instructions, see:

Verify your connection

Before running the examples, verify that the MCP Server is connected and its tools are available.

GitHub Copilot CLI

  1. Start GitHub Copilot CLI.
  2. Enter /mcp.
  3. Check that the Semaphore MCP Server is connected.

Visual Studio Code with GitHub Copilot Chat

  1. Open VS Code with the Semaphore MCP Server configured.
  2. Start the MCP server connection.
  3. Verify that the server status is Running.
  4. Open GitHub Copilot Chat and select Agent mode.
  5. Confirm that the Semaphore MCP tools are available.

Example 1: List available knowledge models

Start by finding the knowledge models you can access.

Enter the following prompt:

List all knowledge models available in Semaphore.
Show each model's name and URI.
Do not modify any data.

The client can use getAllModelsIndexedByURI to retrieve the available models.

Expected result: A list of model names and URIs.

You can use a model URI from the response in subsequent prompts.

Example 2: Count concepts in a model

After identifying a model, you can find out how many concepts it contains.

How many concepts are in the model named "My Model"?
Show the model URI and the concept count.
Do not modify the model.

Replace My Model with the name of your knowledge model.

The client can use getModelUri to identify the model and getConceptCount to retrieve its concept count.

Expected result: The total number of concepts in the selected model.

Example 3: Find concepts by label

You can search for concepts using their preferred labels.

Find concepts whose preferred labels contain "energy"
in the model named "My Model".

Show the matching concepts, their preferred labels, and their URIs. Do not modify any data.

Replace My Model with your model name and energy with the text you want to search for.

The client can use getConceptsByLabelMatch.

This tool supports two matching options:

  • startsWith: Case-insensitive matching.
  • contains: Case-sensitive matching.

Expected result: A list of matching concepts, including their labels and URIs.

If no concepts match, try another search term or matching option.

Example 4: Inspect a concept and its labels

After finding a concept, you can retrieve its details and labels.

Find the concept named "Solar energy" in the model
"My Model".

Retrieve the concept's available properties, preferred labels, and alternative labels.

Show the concept URI and the values returned by Semaphore. Do not modify any data.

The client can use:

  • getConceptsByLabelMatch to find the concept.
  • getConceptAsJsonByUri to retrieve its properties.
  • getLabelsForConcept to retrieve its labels.

Expected result: The concept URI, available concept properties, and associated labels.

For example, if a concept has the preferred label Solar energy and the alternative label Solar power, both labels can be returned.

Note: Concept retrieval tools may return only a subset of the concept's properties. Metadata values, such as skos:definition, may not appear in the response even when they are present in KMM.

To inspect current metadata values, open the concept in Knowledge Model Management and review its Metadata section.

Example 5: Retrieve concepts from a larger model

When working with a model containing many concepts, retrieve the data in smaller batches.

Retrieve the concepts from the model "My Model"
using pagination.

Start with the first page and show each concept's URI and preferred label.

Do not modify the model.

The client can use:

  • getConceptCount to determine the model size.
  • getConceptAsJsonsPaginated to retrieve concepts in pages.

Expected result: A page of concepts and their available properties.

Pagination helps reduce the amount of data returned in a single response and avoids exceeding client token limits.

Example 6: Inspect metadata types

Before adding or updating metadata, identify the metadata types available in the model.

List the metadata types available in the model
"My Model".

Show each metadata type's name, URI, and supported data type information.

Do not create or modify metadata.

The client can use getAllMetadataTypes.

Expected result: Metadata type definitions, including their URIs and available schema information.

Important: getAllMetadataTypes returns metadata type definitions, not the metadata values assigned to individual concepts.

Example 7: Review model changes

You can ask the MCP client to retrieve changes recorded in a model's history.

Show the changes recorded in the model "My Model"
since 1 November 2026.

Summarize the changes and identify the affected concepts and properties.

Do not modify any data.

The client can use getListOfChangesInModelSince.

Expected result: Changes recorded since the specified date, subject to the available history.

For example, the history may include the addition of a concept definition or an alternative label.

Note: Change history is not the same as the current state of a model. A value recorded as added may have been modified or removed later.

Example 8: Prepare changes safely

The MCP Server also provides tools for creating and modifying KMM content.

Before asking an AI client to make changes, identify the target model and confirm the intended operations.

For example:

I want to add three concepts to the model "My Model":

    
  • Solar energy
  • Wind energy
  • Hydropower

First, inspect the model and its available concept classes. Explain which MCP tools you would use.

Do not create or modify anything until I approve the proposed changes.

The client can use read-only tools such as getAllModelsIndexedByURI and getAllCurrentClasses to prepare the operation.

After reviewing the proposal, you can authorize the client to perform the supported modification operations.

For changes that require review before they are applied to the master model, consider using a KMM task.

Important: Some MCP tools modify the master model directly. Always verify the target model or task before approving write operations.

Other available operations

The Semaphore MCP Server exposes tools for:

  • Creating and modifying concepts.
  • Managing concept classes and schemes.
  • Adding and updating labels.
  • Managing relationships between concepts.
  • Creating and updating metadata.
  • Working with tasks and committing changes.
  • Importing RDF triples.

For the complete list of tools and their supported operations, see Available MCP Tools.

Alert