Use the --query parameter to execute a JMESPath query on the results of your az command. For example, the following lists the sql endpoints for each Azure cloud.

az cloud list \
  --query [*].[name,endpoints.sqlManagement]

[
  [
    "AzureCloud",
    "https://management.core.windows.net:8443/"
  ],
  [
    "AzureChinaCloud",
    "https://management.core.chinacloudapi.cn:8443/"
  ],
  [
    "AzureUSGovernment",
    "https://management.core.usgovcloudapi.net:8443/"
  ],
  [
    "AzureGermanCloud",
    "https://management.core.cloudapi.de:8443/"
  ]
]

Get the hostname of a specific web app. This query also formats the output as tab-separated values, which is useful when you want to assign the output to a variable in your script.

az appservice web show \
  -g myRg \
  -n myApp \
  --query hostNames --out tsv

Get then name and managed disk ID, again formatted as tab-separated values. The query labels the name property "name" and it labels the managed disk id "md_id".

az vm list \
  --query "[].{ name:name, md_id:storageProfile.osDisk.managedDisk.id }" -o tsv

Use the JMESPath contains function to select objects. In this case, select the VMs in a specific resource group.

az vm list \
  --query "[?contains(resourceGroup,'myRg')].{ resource: resourceGroup, name: name }"

In this case, select the VMs that have the vmSize 'Standard_DS1'.

az vm list \
  --query "[?contains(hardwareProfile.vmSize, 'Standard_DS1')]"

Use the --out parameter to format the output as tab-separated values and pipe that through grep.

az vm list \
  --query "[].{ name: name, os: storageProfile.osDisk.osType }" \
  --out tsv  \
| grep Linux

You can pipe the command output to JMESPath-terminal and experiment with your JMESPath query there.

pip install jmespath-terminal
az vm list | jpterm

There is a good tutorial for JMESPath at JMESPath.org/tutorial.