Apply role restrictions
Configure which roles may execute a tool. An empty role list leaves the tool unrestricted by role; administrators retain access.
MCP Gateway
MCP Gateway: give AI useful tools.
Keep control of the actions.
Imperum MCP Gateway connects AI assistants and agents to your security tools. Publish the capabilities they can call, apply execution policies and review the outcome—while Imperum manages the connected services’ credentials.
Publish selected toolsControl executionReview outcomes
From AI request to governed action
Explore a request
01 / MCP clientExternal AI assistantAuthenticated caller
separate from service credentials
Search recent alerts
Gateway accessquery: suspicious sign-inseparate from service credentials
02 / Policy checksMCP Gateway
Input schemaValidate
Caller rolesAuthorize
Rate policyCheck
Risk & approvalAllowed
Service credentials managed by Imperum
03 / Tool & outcome
Imperum alertsPublished platform tool
Alert context returned
Suspicious sign-in · workstation-17
SuccessTools behind the gatewayImperum capabilities·Connected services·Registered MCP servers
Execution Log
search_alertsExternal callerSuccessThe result returns to the caller. The execution record identifies the tool, caller and outcome.
01 - Publish
Turn connected capabilities into useful AI tools.
Discover actions from deployed connectors, Imperum’s platform tools and AI services. Choose what to publish and inspect the input schema—the fields an AI must provide to call the tool correctly.
Set risk, approval and rate policies around that capability. Your team can reuse existing integrations without handing each AI client a separate set of service credentials.
A tool with a clear contract
Search Alerts
Publishedsearch_alerts- Source
- Imperum platform
- Input schema
- Search query and filters
- Risk
- Low
- Approval
- Not required in this example
02 - Execution policy
Let routine work through. Put a boundary around risk.
The gateway checks a published tool’s inputs, the caller’s roles, the rate policy and the approval requirement before execution. An allowed call reaches the tool; a denied call stops, and a request needing approval waits.
Set call limits
Use per-tool rate policies to govern request volume. The limit applies to the tenant, tool and caller.
Require approval
High and critical risk require approval by default. Deployed agents can use an administrator-configured auto-approval policy within its allowed risk level.
Permission to connect is not permission to execute. Published-tool discovery and execution checks are separate. Review the policies before enabling an AI workflow.
03 - Server and client
Bring your AI. Bring your tools.
MCP—the Model Context Protocol—gives AI systems a structured way to discover and call tools. Imperum works on both sides of that connection.
Imperum as the server
Give an external AI useful security context.
External MCP clientImperum MCP GatewayPublished Imperum & connector tools
An authenticated MCP client can request published tools. Execution follows the gateway’s policies, so your existing connections become reusable capabilities for AI workflows.
Imperum as the client
Use tools from a server you already run.
Configured Imperum AI workflowGateway & external MCP clientYour registered MCP server
Register the server and discover its tools. Newly discovered tools enter the gateway with high risk and approval required; review their classification and policy before use.
Agent Studio workflows can use gateway tools. AI Assistant access must be enabled and scoped. These settings do not mean every native Imperum agent runs through the MCP Gateway.
Explore Agentic AI Studio04 - Execution review
See what ran. Understand what stopped.
The Execution Log shows caller, tool, status and duration. Review successful executions, failures, policy denials and requests waiting for approval, so the team can investigate an outcome without relying on the AI’s answer alone.
Execution Log
ToolCallerOutcome
Search AlertsAlert context returned
External AI clientSuccessIsolate hostAction has not executed
External AI clientPending approvalConnector actionCaller role not permitted
External AI clientDeniedGetting started
Start with one useful workflow.
Choose a real task, such as gathering alert context. Establish the tool access and review the result before adding actions with greater impact.
- 01Connect and select
Use a configured integration or register an external MCP server. Review the tools exposed to the gateway.
- 02Set the boundaries
Inspect schemas, roles and risk. Decide where approval is needed and configure the caller’s access.
- 03Test and review
Validate a representative request, inspect its outcome and review the execution record before expanding access.
Questions about MCP Gateway.
1Does the AI receive our service credentials?
The caller authenticates to the gateway. Imperum uses managed credentials for connected services and stores registered MCP server secrets encrypted in its vault. Those services still receive the authentication needed for their own connection.
2Does connecting expose every platform action?
No. The gateway exposes published tools. Administrators select local capabilities; discovery from a registered external MCP server adds its tools with high risk and approval required. Tool execution remains subject to the configured policies.
3Which actions wait for human approval?
Tools marked as requiring approval, and high or critical risk tools, wait by default. A deployed agent’s administrator-configured auto-approval policy can permit actions within its risk level. Approval is a policy decision, not a universal step after every tool call.
4Is MCP Gateway the same as Virtus Sentinel?
No. MCP Gateway governs tool calls and connects AI workflows to capabilities. Virtus Sentinel addresses AI asset discovery and exposure. A gateway connection does not by itself provide all of Sentinel’s protections.
Explore Virtus SentinelHave any other questions? Talk to our team
Connect one AI workflow.Define what it can do.
Bring a security task and the tools behind it. We’ll show you how to connect the workflow and set its execution boundaries.