MCP
The Intelligence Hub acts as an MCP Server, exposing MCP Services that AI Agents can use to access industrial data, execute operations and perform configuration and troubleshooting. This section covers how to enable MCP, and the different MCP Tools the hub exposes.
Enable the MCP Server
The MCP server can be enabled by navigating to the Settings page and enabling the server under the MCP section.
Clicking the “Disabled” button on the MCP page will take you directly to the MCP section of the Settings page.
The MCP server supports the Streamable HTTP protocol, and adheres to the 2025-03-26 revision of the MCP specification.

MCP Services
The MCP Services page provides a view of all the MCP Tools configured in the project. Tools that are active and enabled are shown in green. Tools that are disabled are shown as grayed out. Clicking a tool shows the tool details.
Custom Tools
The Custom Tools tab allows you to build custom MCP tools that can access industrial data or perform operations. Each custom tool calls a pipeline when an AI Agent uses the tool. The tool name, description, and parameters are included as context to the AI Agent.
To create a custom tool, click the + button in the MCP Services panel, configure the tool details, and click Save.

Tools can be exported to JSON and imported into another hub using the Export and Import buttons on the MCP Services toolbar. You can export all tools or a subset, then import them into another hub and wire each tool to that hub’s pipelines.
Tool Details
| Setting | Description |
|---|---|
| Tool Name | The name of the tool as presented to the AI Agent. |
| Description | Describes what the tool does and when to use it. The description is sent to the AI Agent along with the tool name and parameters. See Tool Descriptions. |
| Parameters | The payload structure the AI Agent sends when it calls the tool. Each parameter includes a name, type, and optional default value. Parameters without default values must be provided by the AI Agent. If the payload doesn’t match the defined parameters, the tool call fails. Parameter values are passed to the pipeline as the event value. |
| Pipeline To Call | The pipeline that runs when the AI Agent calls the tool. The pipeline must include a Return stage to return a response to the AI Agent. Multiple tools can call the same pipeline, for example to expose the same logic with different names, descriptions, or default parameter values. |
| Group As | The folder the tool is shown under in the MCP Services panel. Grouping is for organization only and doesn’t change how the tool is presented to the AI Agent. |
| Tags | Tags assigned to the tool. Tags can be used with roles and claims to control which tools an AI Agent can access. See Access Control. |
| Enabled | Only enabled tools are exposed to AI Agents. Disabled tools are shown as grayed out in the MCP Services panel. |
Tool Descriptions
Think of the description field as part of prompt engineering, where you’re instructing the LLM when and how to best use a tool. As an example, a tool description like “call this to create a work order” is very little context. But something more verbose like “call this to create a maintenance work order when you think work needs to be performed on a machine. Make sure you always include the machineID of the machine that needs the work and a verbose message on what work needs to be done and why you think it needs to be done” will likely yield better results. Prompt engineering is not a science, but generally the more context the better.
.SystemTrigger with the source set to MCP.Internal Tools
The Internal Tools tab shows the built-in tools the Intelligence Hub provides for connections and project configuration.

Connections
Currently OPC UA, MQTT support out-of-the-box tools for basic data discovery and read-only operations. These tools are generally not meant
to be used in production, as they provide broad and read-only access to the data sources. However they are useful when using a chat bot or AI Agent
to ask questions of the data. To enable these tools, navigate to the Connection Settings for the connection and enable the Enable MCP Tools option.
The MCP Client Connection also supports the Enable MCP Tools option. When enabled, this connection includes all the tools provided by the underlying MCP Server it’s connected to.
Configuration
Configuration tools let AI Agents build and edit project configuration, and explore how configured project entities relate to each other.
Configuration tools are currently not shown in the MCP Services tree. They can be enabled by going to the Settings page and enabling the Configuration Tools option under the MCP section.
CRUD Tools
Create, read, and update pipelines, connections and their inputs/outputs, models, instances, and namespace nodes.
| Tool | Actions | Description |
|---|---|---|
config_pipeline | list, get, create, update | Manage pipelines |
config_connection | list, get, create, update | Manage connections |
config_connection_io | list, get, create, update | Manage inputs/outputs on a connection |
config_model | list, get, create, update | Manage models |
config_instance | list, get, create, update | Manage instances |
config_namespace | list, get, create, update, move | Manage namespace nodes |
Schema Tools
Look up stage, trigger, and connector schemas for use when building pipelines and connections.
| Tool | Actions | Description |
|---|---|---|
config_pipeline_schema | get_pipeline_outline, list_stages, get_stages | Get the pipeline schema (without stage and trigger types), list all stages and triggers, and get stage and trigger schema |
config_connection_schema | list, get, get_input, get_output | List all connection types. Get connection, input, and output schema |
config_model_instance_schema | get_instance, get_model | Get the model and instance schema |
config_namespace_schema | (none) | Get the namespace node schema, covering both local and remote nodes |
Graph Tools
Read-only exploration of how configured entities (pipelines, connections, inputs/outputs) relate to each other.
| Tool | Actions | Description |
|---|---|---|
graph_schema | describe, get_schema | Semantic description of the graph model, or its live structural schema |
graph_query | cypher, sql | Run a read-only query against the project graph |
Address space tools
Helper for browsing existing sources/targets to reference in a pipeline’s Read/Write stages.
| Tool | Description |
|---|---|
config_reference_browse | Browse the project catalog (connections, instances, pipelines, system data, etc.) for use in Read/Write stage configuration |
config_test_read | Test read data sources. Options include existing_input, existing_instance, dynamic_input, and dynamic_instance. The dynamic reads allow the agent to test read inputs and instances that are not part of the project. |
Connecting MCP Inspector
MCP Inspector is an open source tool used to connect to an MCP Server and view tools. To test using MCP Inspector do the following:
- Follow the instructions here to launch MCP Inspector https://modelcontextprotocol.io/docs/tools/inspector
- Change the Transport Type to Streamable HTTP
- Change the URL to point at the host and port configured in the
HTTP Serversection of the Settings page (http://localhost:8885 for the default host and port for our HTTP Server) - Append
/mcpto the URL, after the port - Expand Authentication and copy in the bearer token created (see Server Authentication) for details on creating the token
- Click Connect
- Click List Tools
- You should see the enabled custom tools the user is permitted to use, and the parameters for each tool

Tool Refresh
The MCP server serves its tool list from a cache that it refreshes in the background on a configurable interval (the Tool Refresh Period setting, default 5 minutes) and immediately whenever a connection is saved. Connected clients won’t see those changes until they re-request the tool list, which you can generally force by one of the following:
- Restart the MCP server (toggle off/on). This will force clients to reconnect and re-request tools
- Restart the MCP Client (or agent)
- Prompt the agent/LLM to refresh
MCP does support a way to notify clients of tool changes, but this currently isn’t supported given that most clients don’t support it either. In general, tool changes only occur during development, and should be rare in production workflows.
Access Control
The MCP Server uses the existing roles and claims system to manage agent access.
The following table lists the MCP claim components required for each type of tool category. Use these when assigning roles/claims to users or API keys.
| Tool Category | Resource | Actions |
|---|---|---|
| Custom | mcp tool | agent use |
| Connection | connection | read, execute read, execute write |
| Config | pipeline | read, create, update |
| Namespace Config | namespace | read, create, update |
Custom Tool Permissions
Access to custom tools is controlled by the mcp tool resource. The agent use action controls which custom tools an MCP client
can list and call over MCP. A custom tool is exposed to an MCP client only when the tool is enabled and the user has agent use permission for it.
agent useis the only action required to list and call a custom tool over MCP. It doesn’t require thereadorlistactions onmcp tool.- When a permitted tool is called, its pipeline runs regardless of the user’s
pipelinepermissions. Grantingagent useon a tool effectively grants permission to run the pipeline behind it. - The
create,read,list,update, anddeleteactions onmcp toolcontrol managing custom tools in the MCP Services page and the REST API. They don’t affect which tools are exposed over MCP. - Add a Contains Tag scope to the claim to limit
agent useto tools with specific tags. This lets you give different agents access to different sets of custom tools.
User and API Key Setup
A new user can be created and assigned the appropriate roles and claims for MCP access. An API Key can be generated for this user, which can be used as a Bearer token to authenticate requests to the MCP Server. See REST Data Server for details on creating the API Key.
Additional information on managing users, roles, and claims is available in the Users and Roles documentation.
Examples
Scope custom tools per agent
Suppose you have two agents, a maintenance agent and a quality agent, and each should only see the custom tools built for it.
- Create a
Maintenancetag and aQualitytag (see Tagging for details) - Assign the tags to the custom tools, either in each tool’s Tags field or in bulk from the Tags page
- Create a role for each agent with a claim on the
mcp toolresource, theagent useaction, and a Contains Tag scope with that agent’s tag - Assign each role to a separate user, and generate an API key for each user
- Configure each agent to use its own API key as the Bearer Token. Each agent only lists and calls the custom tools that have its tag
Restrict access to a connection
Suppose you want an agent to only have access to a specific MQTT Connection you configured. Tag-based access control enables this restriction:
- Create a tag (see Tagging for details) and assign it to the specific MQTT connection
- Define a role with a claim that specifies the required resources, actions, and tags
- Assign the role to a user
- Generate an API key for the user and use it as a Bearer Token for MCP Client authentication against the HighByte MCP Server
