View AI Governance logs for MCP Inbound tool calls - The Control Room logs every tool call that a third-party AI assistant makes to the Automation Anywhere MCP server, giving Automation Admins centralized visibility into inputs, outputs, timestamps, and execution details for governance, compliance, and troubleshooting of agentic workflows.

Automation 360

The Control Room logs every tool call that a third-party AI assistant makes to the Automation Anywhere MCP server, giving Automation Admins centralized visibility into inputs, outputs, timestamps, and execution details for governance, compliance, and troubleshooting of agentic workflows.

About this task

When a third-party AI assistant such as Microsoft Co-pilot, SAP Joule, or any external agent calls a tool exposed by the Automation Anywhere MCP server, the Control Room automatically captures a governance log entry. All such entries appear in the Agents tab of AI governance logs and are prefixed with MCP Inbound to distinguish them from native AI Agent activity.

Tool calls that originate from the same AI assistant session are chained together in a single log entry, giving you an end-to-end view of the full interaction. Each entry records the following details:

  • Tool called and the time it was called
  • Input variable names, data types, and the values passed to the tool
  • The output (response) returned by the MCP server
  • Execution status, run-as user, session ID, and device name

This governance capability supports enterprise audit and compliance requirements while enabling authorized users to troubleshoot agentic workflows. The security focus of MCP Inbound governance ensures that all external AI interactions are transparent and attributable.

Before you begin

Ensure you have the Automation Admin role or equivalent permissions to access AI Governance logs in the Control Room. At least one MCP Inbound tool call must have occurred for entries to appear in the log. To review or configure the MCP tools available to external AI assistants, see Configure unattended automations on remote devices with MCP inbound.

Procedure

  1. Log in to the Control Room and navigate to AI > Governance.
  2. On the AI governance logs page, select the Agents tab.

    The Agents tab displays detailed logs for both native AI Agent interactions and third-party agent interactions with the MCP server. MCP Inbound entries are listed with the following columns:

    Column Description
    Status Current state of the tool call activity, such as Active or Completed.
    Activity name Name of the tool call, prefixed with MCP Inbound followed by the tool name (for example, MCP Inbound - Create_Workday_Request). Click the name to open the detailed log view.
    Agent name Displays External Agent for tool calls originating from a third-party AI assistant.
    Source Displays External AI Agent for MCP Inbound entries.
    Running time Duration of the tool call execution.
  3. Optional: To narrow the results, use the Time filter drop-down to select a time range, or use the Activity name search field to locate a specific tool call.
  4. Click an activity name to open the detailed log view for that MCP Inbound session.

    The detail view contains two panels:

    • Agent activities (left panel): Lists all tool calls chained within the same AI assistant session. Each call is identified by a reference number, such as Ref 1-1 and Ref 1-2, and shows the start time, running time, and status. Tool calls are chained here when they originate from the same external AI assistant, giving you a full sequence view of the agentic interaction.
    • General (right tab): Displays summary information about the overall session.
  5. In the Agent activities panel, click a specific tool call to review its full execution details.

    The detail panel on the right shows the following sections:

    Input
    A table showing the input variable name, data type, and the value that the AI assistant passed to the tool when making the call. For example, a Create_Workday_Request call might show a Title variable of type String and a Description variable of type String.
    Output
    A table showing the output variable name, data type, and the response value returned by the MCP server after the automation executed.
    Run details
    Execution metadata including the run-as user, session ID, and the name of the device on which the automation ran.
  6. Optional: To export governance log data, click Export Agents to .csv in the upper-right corner of the AI governance logs page.

Results

You now have a complete audit trail of MCP Inbound tool calls for the selected session, including the inputs sent by the AI assistant, the MCP server responses, and execution metadata. Chained tool calls from the same AI assistant session are consolidated in a single entry, providing end-to-end traceability for each agentic workflow.

What to do next

To manage the MCP tools that external AI assistants can discover and call, navigate to AI > Agent connections in the Control Room. Each connection listed on this page corresponds to a tool that third-party AI assistants can invoke through the MCP server.