AppViewX ADC MCP Server

The AppViewX ADC MCP Server enables MCP-compatible AI assistants to query Application Delivery Controller (ADC) operational information from the AppViewX Platform using approved read-only tools.

This release introduces natural language access to ADC operational visibility use cases, including device inventory queries, ADC object retrieval, topology discovery, and operational health monitoring all through a standardized Model Context Protocol (MCP) interface.

With the ADC MCP tools, ADC administrators, network operations engineers, and IT teams can retrieve device and object information directly from their preferred AI-assisted workflow (VS Code with GitHub Copilot or Claude Desktop) without navigating the AppViewX UI.

Key Capabilities

  • Query ADC device inventory details using natural language prompts.
  • Retrieve ADC object information including virtual servers, pools, monitors, and more.
  • Filter results by vendor, status, environment, device group, object type, and other criteria.
  • Receive structured, human-readable responses with operational metadata.
  • Operate within RBAC-controlled access boundaries — users see only what they are authorized to access.

How It Works

The ADC MCP Server acts as a bridge between your MCP-compatible AI assistant and the AppViewX Platform. The communication follows a secure, standardized flow:

  1. You ask a question — Type an ADC-related query in your MCP-compatible AI assistant (e.g., Show disabled virtual servers).
  2. The AI assistant identifies the right tool — Based on your query, the AI assistant automatically determines which ADC MCP tool to invoke (device retrieval or object retrieval).
  3. MCP tool call is sent — The AI assistant converts your request into a structured MCP tool call and sends it to the AppViewX ADC MCP Server over stdio.
  4. Authenticated request to AppViewX — The MCP Server translates the tool call into an authenticated HTTPS request to the AppViewX Platform, respecting RBAC permissions.
  5. Data is returned — The AppViewX Platform processes the query against the ADC inventory and returns the matching results.
  6. Human-readable response — The MCP Server formats the response and sends it back to the AI assistant, which displays the results in a table, bulleted list, or summary format.
Note: The ADC MCP tools are read-only. They expose operational information for visibility and audit purposes. Configuration changes, device modifications, or execution workflows are not supported through MCP.

What's Available

The AppViewX ADC MCP Server provides two tools for querying ADC operational information:

Tool Description Input / Filter By Returns
ADC Device Retrieval

Retrieve device inventory details for ADC devices configured in AppViewX.

 Query and filter the ADC device inventory to retrieve device profiles, operational status, platform metadata, and health metrics.

Filter by: name, ip, fqdn, dataCenter, vendor, status, version, or free-text keyword search.

The AI maps natural language to the correct filters automatically. Supports pagination with max and start parameters.

 Matching device list with metadata including name, IP, FQDN, device type, module, data center, vendor, status, LB sync, onboarding group, certificate management, host device, HA active status, version, management IP, sync group, expiry date, and tenant list.
ADC Object Retrieval

Retrieve ADC object details including virtual servers, pools, monitors, and topology metadata.

 Query and filter ADC objects across your device inventory to retrieve configuration metadata, operational state, and application associations.

Filter by: name, ip, fqdn, parentName, code, partition, status, deviceName, vendor, isOrphan, or free-text keyword search.

Supports pagination with max and start parameters.

Supported ADC object types: Wide IP, GTM pool, GTM pool member, virtual server, LTM pool, LTM pool member, profiles, iRules, policies, monitors, DataScript, IPGroup, DNS loadbalancer, HTTP/TCP load balancer.

 Matching object list with metadata including name, IP, FQDN, partition, object type, device, device IP, vendor, state, status, domain, tenant, namespace, record type, and last status sync time.
Tip: Both tools are available in a single MCP session. You can ask device-related and object-related questions in the same conversation. The AI assistant automatically routes your query to the appropriate tool.

Getting Started

Follow these steps to set up and configure the AppViewX ADC MCP Server:

  1. Verify that all prerequisites are met.
  2. Create a service account in AppViewX.
  3. Download and extract the ADC MCP Server.
  4. Configure your AI client (Claude Desktop or VS Code).
  5. Verify the connection.

Prerequisites

Requirement Version Notes
AppViewX Platform 2026.3.0+ Must be reachable over HTTPS from your local machine.
Java Runtime Environment Latest Install the latest supported Java version and ensure Java is available on your system PATH. Verify the installation using java -version.
AI Client Latest Use Claude Desktop or Visual Studio Code with the GitHub Copilot extension.
AVX MCP JAR Latest Download the avx-mcp JAR file from the Agents and Downloads tab on the AppViewX platform. See step 2.
AppViewX Service Account — Create or use an existing service account to obtain the Client ID and Client Secret required to authenticate with AppViewX. See step 1.
ADC Module Access — Ensure the service account is mapped to a user group that has the required ADC permissions.

Step 1: Create a Service Account in AppViewX

A service account provides secure, non-interactive credentials for the MCP Server to authenticate with the AppViewX Platform.

  1. Log in to your AppViewX instance.
  2. Navigate to Platform > Service Account.
  3. Click Create New Service Account.
  4. Enter a descriptive name for the account.

    Example: adc-mcp-integration

  5. Under Role, assign the built-in ADC MCP read-only role.
    Note: AppViewX exports a default role specifically designed for MCP framework access. This role contains the required ACL permissions to access the ADC MCP tools. Assign this role to ensure minimum-privilege access.
  6. Map the service account to a user group that has the appropriate ADC permissions.
    • Create a new user group with the ADC MCP role, or use an existing group with ADC visibility permissions.
    • The user group's ACL determines which devices and objects are visible through the MCP tools.
  7. Under Resources, attach the device groups or environments that this service account needs to access. Limit the scope to only the resources required.
  8. Copy and securely store the generated Client ID and Client Secret.
    Important: The Client Secret is shown only once during creation. Store it securely. If lost, you must regenerate the credentials from the Service Accounts page.

Step 2: Download and Extract the ADC MCP Server

  1. Log in to AppViewX and click the (?) icon in the top-right corner.
  2. Select Agents and Downloads.
  3. Locate the AppViewX ADC MCP Server widget and click Download.
    The file avx-adc-mcp-<version>.zip is saved to your Downloads folder.
  4. Extract the ZIP to a permanent location.
    Tip: Choose a stable path — for example, C:\mcp. The full path to the JAR file will be referenced in the configuration step and must not change after setup.

    After extraction, the folder contains:

    avx-adc-mcp-26.3.0.0/
├── lib/
│   └── avx-adc-mcp.jar    ← the server executable
├── mcp.json                ← sample client configuration
└── README.md
  5. Note the full absolute path to avx-adc-mcp.jar.

    Example (Windows):

    C:\mcp\avx-adc-mcp-26.3.0.0\lib\avx-adc-mcp.jar

    You will need this path in the next step.

Step 3: Configure Your AI Client

Choose the option that matches your AI client.

Option A: Configure Claude Desktop

  1. Open Claude Desktop.
  2. Navigate to Settings > Developer > Edit Config.
    The file claude_desktop_config.json opens in your default text editor.
  3. Add the avx-adc-mcp-server block inside the mcpServers object: 
    {
  "mcpServers": {
    "avx-adc-mcp-server": {
      "type": "stdio",
      "command": "java",
      "args": [
        "-jar",
        "C:\\mcp\\avx-adc-mcp-26.3.0.0\\lib\\avx-adc-mcp.jar"
      ],
      "env": {
        "AVX_BASE_URL": "https://your-appviewx.example.com:31443",
        "CLIENT_ID": "your-client-id",
        "CLIENT_SECRET": "your-client-secret"
      }
    }
  }
}

    Replace the placeholder values:

    Parameter Description
    JAR path The absolute path to avx-adc-mcp.jar on your machine. Use double backslashes \\ on Windows.
    AVX_BASE_URL Your AppViewX instance URL including port (e.g., https://appviewx.company.com:31443).
    CLIENT_ID The Client ID from your AppViewX service account (Step 1).
    CLIENT_SECRET The Client Secret from your AppViewX service account (Step 1).
  4. Save the file and fully restart Claude Desktop.
    Claude Desktop spawns the ADC MCP Server process automatically on startup. AppViewX ADC tools are now natively available inside Claude.

Option B: Configure VS Code (GitHub Copilot)

  1. Open VS Code.
  2. Press Ctrl+Shift+P to open the Command Palette.
  3. Type and select MCP: Open User MCP Configuration.
    Tip: To limit this server to a single project, select MCP: Open Workspace Folder MCP Configuration instead. This creates .vscode/mcp.json in your current workspace folder only.
  4. Add the avx-adc-mcp-server entry into the servers object: 
    {
  "servers": {
    "avx-adc-mcp-server": {
      "type": "stdio",
      "command": "java",
      "args": [
        "-jar",
        "C:\\mcp\\avx-adc-mcp-26.3.0.0\\lib\\avx-adc-mcp.jar"
      ],
      "env": {
        "AVX_BASE_URL": "https://your-appviewx.example.com:31443",
        "CLIENT_ID": "your-client-id",
        "CLIENT_SECRET": "your-client-secret"
      }
    }
  }
}

    Replace the JAR path, AVX_BASE_URL, CLIENT_ID, and CLIENT_SECRET with your actual values.

  5. Save the file (Ctrl+S).
    VS Code detects the configuration and automatically starts the ADC MCP Server process.
Important: Always use absolute paths — do not rely on working directory resolution. Ensure Java is available in PATH before starting the AI client. Avoid wrapping the JAR in shell scripts unless strictly necessary.

Step 4: Verify the Connection

Confirm the AppViewX ADC MCP Server is running and tools are available to your AI assistant.
Verify the connection in your AI client.
In Claude Desktop:
  • Check that the ADC tools appear in the available tools panel.
  • Send the test prompt: Show me available tools in AppViewX ADC MCP.
  • Claude lists the available ADC tools (device retrieval and object retrieval) if the connection is successful.
In VS Code:
  • Press Ctrl+Shift+P and select MCP: Show Installed Servers.
  • The server avx-adc-mcp-server appears with a status of Running or Started.
  • Two ADC tools (device retrieval and object retrieval) are visible beneath the server entry.
The AppViewX ADC MCP Server is now ready. You can query your ADC device inventory, retrieve object details, check device health, and explore topology information through natural language conversation.

Troubleshooting

Common issues and resolutions when setting up or running the AppViewX ADC MCP Server.

Common Issues and Resolutions

Issue Cause Resolution
Server fails to start / JAR not found The path to avx-adc-mcp.jar in the configuration is incorrect.

Double-check the absolute path in your configuration file.

On Windows, verify the file exists:

Test-Path "C:\mcp\avx-adc-mcp-26.3.0.0\lib\avx-adc-mcp.jar"

The command must return True.

Use double backslashes \\ or forward slashes / in the JSON path string.
Java not found error Java is not installed or not on the system PATH.
  1. Install Java 21.
  2. Restart your terminal or AI client after installation.
  3. Verify: java -version
Connection refused The AppViewX URL is incorrect or unreachable from your machine.
  1. Verify AVX_BASE_URL is correct and includes the port.
  2. Test connectivity: 
    curl https://your-appviewx.example.com:31443
  3. Check network/firewall rules between your machine and AppViewX.
Unauthorized / 401 error The Client ID or Client Secret is incorrect or expired.
  1. Log in to AppViewX and navigate to Platform > Service Accounts.
  2. Confirm or regenerate the credentials for your service account.
  3. Update CLIENT_ID and CLIENT_SECRET in your configuration and restart.
403 Forbidden / No data returned The service account lacks the required RBAC permissions for ADC resources.
  1. Verify the service account is mapped to a user group with the ADC MCP role.
  2. Ensure the user group has visibility to the requested device groups or environments.
  3. Check that the default ADC MCP read-only role is properly assigned.
Tools not appearing in AI client Configuration error or stale client state.
  1. Validate JSON syntax in the configuration file (use a JSON linter).
  2. Fully restart the AI client (Claude Desktop or VS Code) to force a fresh server spawn.
  3. Run the JAR manually to confirm it starts without errors: 
    java -jar "path\to\avx-adc-mcp.jar"
Query timeout or rate limiting An overly broad query is retrieving too many records.
  1. Refine your query with more specific filters (e.g., add vendor, status, or deviceName).
  2. Use pagination parameters (max, start) for large result sets.
  3. Narrow the scope of your request to a specific environment or device group.
Server status is not "Running" The server process crashed or failed to initialize.
  1. Verify the JAR path is correct.
  2. Check console output for startup errors by running the JAR manually.
  3. Restart the AI client completely.

Diagnostic Steps

If you encounter issues, follow these steps to diagnose the problem:
  1. Verify Java.

    Run java -version and confirm Java 21 is installed.

  2. Test the JAR manually. 
    java -jar "C:\mcp\avx-adc-mcp-26.3.0.0\lib\avx-adc-mcp.jar"

    Review any error output for connectivity or authentication issues.

  3. Validate JSON configuration.

    Paste your configuration file content into a JSON validator to check for syntax errors.

  4. Check network connectivity.

    Ensure your machine can reach the AppViewX instance over HTTPS.

  5. Verify service account.

    Log in to AppViewX and confirm the service account is active with the correct role and group mapping.

Note: All dates and timestamps returned by the ADC MCP tools are in ISO 8601 UTC format. Ensure your AI client or downstream systems handle timezone conversion appropriately if needed.