13. Operational Governance
Operational governance keeps the Hopex MCP Server controlled in customer environments: configuration, rights, schemas, write operations, supervision, and support information.
The MCP Server exposes stable functional capabilities through a technical implementation that may evolve between versions.
13.1. MCP tool evolution
The MCP tool list may evolve between Hopex MCP Server versions. Tools may be renamed, merged, split, or replaced.
Functional capabilities remain the reference: schema introspection, query construction and execution, authorized mutations, relationship exploration, object counting, and access to diagram information.
Any new capability, functional evolution, or feature removal is documented in the release notes of the update that introduces the change.
13.2. Exposed scope control
The exposed scope is defined by the selected schema, Hopex rights and profile, and the MCP module configuration in HAS. This combination determines the objects, relationships, and operations available to the AI client.
Schema or profile changes are treated as scope changes. They can modify what the AI client can query or return, even when the MCP tool list remains unchanged. Changes involving customized metamodel exposure should be reviewed according to section 5.2 Customized metamodel exposure.
13.3. HAS configuration governance
The Hopex MCP module settings in the HAS console define key integration choices: default GraphQL schema, API key mode, mutation authorization, deferred authentication, and technical parameters.
These changes affect the integration, connected AI clients, machine-to-machine scenarios, write operations, and supervision.
13.4. Write-operation governance
Mutations expose write operations on the Hopex repository. Their use depends on the Allow mutation option in the Hopex MCP module settings. This option is required to execute mutations, regardless of the authentication mode.
In API key mode, the x-allow-mutation header complements this configuration. Irreversible deletions follow an explicit confirmation flow before execution. Applied changes remain traced at the level of the affected Hopex objects.
Hopex records mutations under the user linked to the request. HAS supervision identifies the MCP session, user, AI client, and tool.
The customer is responsible for deciding whether mutations are enabled and which AI clients may use them. For update prompts, the customer must ensure that the prompt asks for user validation before any mutation is executed.
When LLM-generated data is used in a mutation, the customer must verify that it is accurate, relevant, and appropriate for the business context. If validation is performed later in Hopex, it remains the customer's responsibility and is not provided by the MCP Server or by Bizzdesign Hopex.
13.5. AI client compatibility
Registration, authentication, headers, secrets, and error handling vary by AI client. Use the AI client documentation for exact syntax and client-side practices.
On the Hopex side, governance covers the elements interpreted by the MCP Server: endpoint, authentication mode, received headers, schema used, Hopex rights, and supervision events.
13.6. Supervision and support
HAS supervision events are the reference point for monitoring MCP Server usage: sessions, AI clients, identified users, executed tools, loaded resources, results, and estimated volumes.
For support, collected information links the symptom observed in the AI client to the Hopex context: MCP session, declared client, repository, profile, schema, authentication mode, tool or resource concerned, and returned structured response.
13.7. Operational practices
Document changes to the Hopex MCP module configuration in HAS.Track MCP tool changes while distinguishing technical tools from functional capabilities.Limit the exposed scope to the required schema, rights, and profiles.Test AI clients after changes to schema, customized metamodel exposure, authentication, headers, or write-operation configuration.Review supervision events and write-operation usage regularly.Keep a support information collection procedure available.