How to use the App Insights Instrumentation agent skill

What the skill makes the agent do

The skill is a set of instructions and reference guides. It does not run a monitoring service of its own. When you ask for telemetry, the agent works inside your project: it reads your code, decides how the app can be instrumented, and makes the changes.

The agent first identifies the app's programming language, application framework and hosting. It reads the source code to make an educated guess and asks you about anything it cannot determine. It must also ask where the app is hosted, for example on a personal computer, in Azure App Service as code or container, or in an Azure Container App.

• ASP.NET Core app hosted in Azure App Service: the agent points you to the auto-instrumentation guide, which needs no code changes.
• Node.js, Python or ASP.NET Core apps otherwise: the agent creates an App Insights resource and edits the app's code.

Install the skill from the ZIP

Download the ZIP for this skill from this page. Then choose one of two install locations and unzip the skill folder there.

You can install it for yourself across projects by unzipping it into ~/.claude/skills/. You can also install it for one project by unzipping it into that project's .claude/skills/ folder.

If you use the Claude apps, you can instead upload the ZIP under Settings → Capabilities → Skills.

Check what you need before you start

The skill's prerequisites say the app must be an ASP.NET Core app or a Node.js app hosted in Azure. You need an Azure subscription and permission to create App Insights resources. The workflow also needs your app code in the workspace.

For the Azure CLI option you need the Azure CLI. If your project already has Bicep templates, the skill prefers adding App Insights to them.

• Azure subscription with access to create App Insights resources
• Azure CLI, for the CLI option, or existing Bicep templates
• Application code in the workspace (ASP.NET Core, Node.js or Python)

Ask for telemetry with a natural request

You do not need a special command. Describe what you want in plain words, and mention the app's language and where it runs if you know it. The skill is meant to trigger when you want telemetry enabled for your web app.

The agent will ask follow-up questions, such as where the app is hosted, and then guide you through the steps.

• Add Application Insights telemetry to this ASP.NET Core app so I can see errors and slow requests.
• Instrument my Node.js web app in Azure to send telemetry to App Insights.
• Help me monitor this web app's health in Azure. It runs in Azure App Service as a container.

Example: adding telemetry to an ASP.NET Core app

For a manual ASP.NET Core setup, the ASP.NET Core guide asks the agent to install the client library and configure Program.cs. The agent adds a using line at the top of the file and adds a call before builder.Build().

After the code change, the app must be deployed for the change to take effect. The connection string of the App Insights resource is then added as an environment variable of the running app. The agent can use the Azure CLI to read that connection string.

• Install the client library with the dotnet add package command shown in the guide.
• Add the using statement for Azure.Monitor.OpenTelemetry.AspNetCore at the top of Program.cs.
• Before builder.Build(), add builder.Services.AddOpenTelemetry().UseAzureMonitor();
• Deploy the app, then set the connection string as an environment variable.

Safety and review before you accept changes

This skill runs scripts and uses the network. It can create Azure resources and edit your application code. Review every change the agent proposes before you accept it, and do not run it against production without review, since live app code and costs can be affected.

The skill's SKILL.md refers to an Azure CLI script and a Bicep example. Those files are not in the file list for this skill, so check the contents of your ZIP before relying on them.

• Confirm the subscription, resource group and app name before any resource is created.
• Use a resource group that makes management easier, such as the one that already holds the hosted app.
• Test in a non-production environment first.