Understanding MCP tools in AI assistant integration

When you connect an AIClosed Artificial Intelligience assistant to Digi Remote Manager using the MCPClosed Model Context Protocol server, the assistant gains access to a structured set of platform capabilities called tools. These tools represent secure, permission-controlled operations that allow the AI assistant to retrieve information, analyze fleet data, and perform supported management tasks on your behalf.

You do not need to configure or manually enable individual tools. MCP-compatible clients automatically discover the tools available to your account based on your APIClosed Application Programming Interface key permissions.

Each tool represents a specific, well-defined platform capability. When questions are asked, the AI assistant translates the request into one or more MCP tool calls. The platform processes the request securely and returns structured data, which the assistant then formats into a natural-language response.

Key characteristics

  • Standards-based: Tools are exposed through the MCP.

  • Permission-aware: Tool access is governed by the Digi Remote Manager user account that created the APIClosed Application Programming Interface key.

  • Account-scoped: Tools operate only within the account and subaccount scope associated with the API key.

  • Secure by design: All tool interactions occur over HTTPSClosed Hypertext Transfer Protocol Secure using authenticated requests

Caution Do not build hard-coded dependencies on specific MCP tool names or assume they will remain static. AI clients automatically discover available tools on each connection, and implementations should rely on dynamic tool discovery rather than fixed identifiers to ensure compatibility as the platform evolves.

The MCP server exposes a comprehensive set of capabilities that generally fall into the following categories listed in the table.

Tool category

Description
Device inventory and queries

Retrieve information about devices, filter fleets by attributes, and analyze device status across groups or regions.

Telemetry and data streams

Access time-series data, historical metrics, and aggregated statistics for device performance and connectivity.

Live device state and diagnostics

Query real-time device state, configuration data, and operational details to support troubleshooting workflows.

Alerts, monitoring, and automations

Review alert configurations, monitor activity, and inspect automation workflows and execution history.

Firmware, jobs, and configuration

Examine firmware versions, deployment activity, configuration templates, and compliance status.

Account, users, and audit data

Access account-level information, user details, and audit history within the bounds of your permissions.

Fleet summaries and analytics

Generate aggregated counts and summaries across large device fleets to support reporting and operational insights.

These categories allow AI assistants to support both conversational use cases (for example, ad-hoc troubleshooting) and structured AI agent workflows (for example, automated triage and reporting).

  1. Tools are automatically discovered.

    Once the MCP connection is established, your AI client discovers available tools automatically. There is no manual registration process.

  1. Permissions determine access.

    The AI assistant can only access tools and data allowed by the API key’s user permissions. If a request fails due to insufficient permissions, adjust the Digi Remote Manager user role and regenerate the API key if necessary.

  1. Tools are structured, not free-form.

    Unlike open-ended text prompts, MCP tools expect structured inputs (such as filters, identifiers, or time ranges). The AI assistant handles this translation for you, but clear, specific questions improve accuracy.

    For example: “List devices offline in the last 24 hours” is more precise than “Show me problems.”

  1. Tools can be combined.

    AI assistants may invoke multiple tools in sequence to answer a single request. For example, a summary question might require retrieving device data and then generating aggregated counts.

  2. Output format varies by client.

    The MCP server returns structured data. How that data is presented (table, paragraph summary, chart, etc.) depends on the AI client.

To get the most value from AI assistant integration:

  • Use specific, well-defined queries.

  • Include filters such as time ranges, groups, or device identifiers when possible.

  • Confirm that the API key has the necessary permissions for operational tasks.

  • Use separate API keys for different environments (development, testing, production).

  • Validate query results before taking action in operational environments.

The MCP server may expand over time to expose additional platform capabilities. Because the integration is standards-based, MCP-compatible clients automatically discover newly available tools without requiring configuration changes.

Summary

MCP tools provide secure, structured access to Digi Remote Manager capabilities for AI assistants and AI agents. They enable natural-language interaction, automated workflows, and large-scale fleet analytics without requiring custom API integration. Access is controlled by your API key permissions, and tool discovery is automatic once the MCP connection is established.