Integrate an MCP server to manage AppDynamics

Learn how to deploy, secure, configure, and use the Splunk AppDynamics MCP Server.

The Splunk AppDynamics Model Context Protocol (MCP) Server is an integration layer that enables AI agents to interact with AppDynamics data and tools in a structured, secure, and permission-aware manner. It acts as a bridge, allowing AI applications to query AppDynamics data through standardized MCP tools.

Alternatively, you can connect to Virtual Appliance's MCP server. The Virtual Appliance exposes two independent Model Context Protocol (MCP) servers. Add both servers to combine platform operations with AppDynamics application and observability data.

Server Purpose Client URL Authentication
Virtual Appliance MCP Operate and diagnose the AppDynamics Virtual Appliance. https://VA-DNS-NAME/va-mcp Bearer VA-MCP-token
AppDynamics MCP Query AppDynamics APM and EUM data. https://VA-DNS-NAME/appdynamics-mcp Bearer AppDynamics-token
For more information, see Connect to an MCP Server to Simplify Splunk AppDynamics Operations.

This guide explains how to grant access, deploy, secure, configure, and use the AppDynamics MCP Server.

Overview

The AppDynamics MCP Server gives MCP-capable AI clients a read-oriented interface to AppDynamics monitoring data. It normalizes Controller and Analytics Events responses, publishes domain guides, and provides reusable prompt templates for guided investigations. The MCP server provides 105 tools, 19 guided prompts, and seven domain guides across seven domains.

Domain Scope
Application Monitoring (APM) Discover applications, tiers, nodes, business transactions, metrics, flowmaps, health rules, dashboards, snapshots, hotspots, and call graphs.
End User Monitoring (EUM) Analyze Browser RUM, Mobile RUM, Synthetic Web and API Monitoring, user journeys, metrics, geography, and network data. Correlates EUM data with APM.
Analytics Events Discover Analytics Events types and live query-result fields. Run bounded, read-only ADQL queries through the external Events Service.
General Search Controller entities, tags, tagged APM metrics, activity events, health incidents, anomalies, RCA, tags, infrastructure-tag, and alerting policies.
Machine Agent Discover machines, health status, built-in and dynamic metrics, learned baselines, assessment and triage. Correlate infrastructure with APM applications and resolve hosts.
Cluster Agent Discover Kubernetes clusters, pods, containers, and cached events. Assess readiness and capacity, inspect log metadata, correlate APM ownership, and resolve physical hosts.
Database Monitoring (DBMon) Discover database servers and collectors. Analyze KPIs, clusters, waits, top queries, blocking, query details, and metrics. Correlate database activity with APM and SIM.

Supported MCP transports

The server supports two MCP transports:

  • streamable-http for a shared HTTP/HTTPS deployment.
  • stdio when an MCP client starts the server as a subprocess.

It uses two independent downstream credential sets:

  • The AppDynamics Controller, authenticated with a request-scoped bearer token for HTTP or a process-managed token or OAuth credential for stdio. For more information, see Required settings
  • The optional Analytics Events Service, authenticated with a separate Events API key and a server-owned Global Account Name.
    Note: Provide your Events credentials to use the Analytics Events tools. Without these credentials, Analytics Events tools cannot function. See Analytics Events Settings.

For more information about MCP, see What is the Model Context Protocol (MCP)?

Prerequisites

Requirement Detail

Docker-compatible container runtime

Docker Engine or Docker Desktop capable of running Linux containers.

See Install Docker Engine.

AppDynamics Controller On-premises Controller 26.7, network-reachable from the MCP server.
Controller API client A bearer token for HTTP or stdio env_bearer, or OAuth credentials for stdio oauth.
Analytics API client

(Optional) For live Analytics Events tools:

  • The Analytics Events service base origin.
  • The applicable Controller global account name, global account name, or both.
  • A separate Events API key with only the required query grants.

For more information about Analytics Events Tools, see Analytics Events tools.

Supported MCP clients The Splunk AppDynamics MCP Server is compatible with MCP-compliant AI clients that support streamable HTTP or stdio transport, such as Claude Desktop and Codex.

System requirements

These are the recommended values for HTTP and STDIO modes. However, you can change the values depending on your tool complexity, response size, Controller performance, and concurrent usage.
HTTP
Replica count CPU request / limit Memory requests / limit Rate Inventory-only workload
1 250m/500m 256mi /512 mi up to 5 tools per second

up to 8 requests/second validated for six minutes

Note: For higher concurrency or burst-heavy use, consider 1 CPU/1Gi per replica or add replicas.
STDIO
Each STDIO client runs a separate MCP server process. For each process, allocate approximately:
CPU Memory Rate Short bursts
500m 256mi 1 tool request per second

Up to 3 requests/second when client call allows concurrent calls.