MCP stands for Model Context Protocol. It is an open standard that allows AI platforms to securely connect with external data sources and systems. The Alfabet MCP Server provides controlled access to data stored in the Alfabet database through a set of predefined MCP tools. When an AI service connects to the MCP server, it retrieves the available tool catalog and determines the most appropriate tool based on the user's request.

The AI service then invokes the selected tool, passing filter criteria derived from the user's input. In response, the MCP server executes related API calls internally to retrieve to retrieve the requested data from Alfabet .

The retrieved information is returned to the AI service in a structured format, enabling it to accurately interpret the data and generate clear, consistent, and well-structured responses to user queries.

Examples for use cases:

  • User request: How many applications are there in the repository for a given status?

    The AI system can understand that the class being referred to is Application in Alfabet . It will filter for applications with the status and do a count. The answer is provided to the user.

  • User Request: Which applications are supporting the Business capability Payroll Processing?

    The AI agent can traverse the Application to Business Capability relationships inside Alfabet using Meta model relationships and provide the answer to the user.

The Alfabet MCP Server provide access to Alfabet data via the following tools:

Search tools handle the majority of natural language questions about enterprise architecture objects. They support wildcards (* for any sequence), property filters composed with AND/OR logic, and offset-based pagination for large result sets.

Tool Functionality
Alfabet search objects Finds enterprise architecture objects by class, property filters, wildcards, and pagination.

For example, a user wants to know how many components reach end of vendor support next year.
Alfabet search multi Searches 2 – 5 classes in parallel. Results are grouped by class.

For example, a user wants a list of all applications and components with a data quality warning the user is responsible for.
Alfabet find person Searches users, which means objects of the object class Person by name, login, or email and returns the REFSTRs property of the users found.

Retrieval tools: Return details about objects or the object count matching specified characteristics.

Tool Functionality
Alfabet get objects by refs If the AI already knows an object's REFSTR object class property value, it can fetch full details directly via this tool.

For example, a user has already received a list of components with end of vendor support next year and requests details about one of the listed components.
Alfabet count objects Returns the number of objects from a single object class matching a specific characteristic. The tool uses a patented binary-search algorithm to count large datasets accurately without downloading thousands of records.

For example, returning how many contracts are ending next year, or how many applications are supporting a business capability.

Graph Traversal tools enable the AI to follow relationships between objects, without needing to know property names in advance.

Tool Functionality
Alfabet get relations Returns all relationships for an object providing the target names. For example, finding the business capabilities an application supports, which projects consume a technology, or who is responsible for a domain.

Meta-model Tools let the AI understand the Alfabet data model itself: what classes exist, what their properties are, how stereotypes map to classes, and what values are valid for enumeration fields. These tools enable the AI to construct correct, targeted queries on the fly. They are prerequisite for deriving answers from the data returned via other tools.

Tool Functionality
Alfabet list classes Lists all object classes and their stereotypes in the metamodel.
Alfabet get class schema Returns all object class properties including information about their data type, and all object class stereotypes for a single object class.
Alfabet resolve class Resolves ambiguous stereotype names to existing stereotypes.
Alfabet get enum values Lists all allowed values for an enumeration.

Report Tools discovers and executes configured reports in Alfabet that have been enabled for REST API access. This gives the AI access to complex pre-built queries — portfolio views, compliance snapshots, lifecycle summaries — without any additional coding. The Applicable for REST API attribute of the configured report must be set to True to make the report available to the MCP server.

Tool Functionality
Alfabet list reports Returns which configured reports are enabled for REST API consumption.
Alfabet run report Executes a named configured report with filter parameters.

Internal information tools inform about the Alfabet MCP server you are currently connected to.

Tool Functionality
Getversion Returns the version of the MCP server.

Operation tools confirm that the server is reachable and force re-authentication if a token has expired mid-session without automatic recovery. The following tools are not listed in the Alfabet MCP Server's list of tools. They are not used for fetching data and are implemented for internal management of the MCP connection.

Tool Functionality
alfabet_check_health Verifies that the Alfabet server and the API components are running.
alfabet_reconnect Forces re-authentication and resets the cached session after token expiry.

AI Clients do not retain Alfabet data between sessions. If a user asks about the same application in different sessions, the AI queries Alfabet independently in each session. This ensures the AI always returns current data.