The Agent Skills Directory

Indirect Prompt Injection Considerations: The skill processes telemetry data, such as exception messages and stack traces, along with Git repository metadata. Since these sources can be influenced by external factors or untrusted inputs, they represent a potential surface where malformed data could attempt to influence the agent's output during the report generation process.
Ingestion points: Untrusted data enters the agent's context through the getAzureMcpEvents_ToolCalls Kusto function and git log metadata as specified in the skill's procedures.
Capability inventory: The skill can execute shell commands via git and generate HTML files to produce reports.
Sanitization: The instructions do not define specific sanitization or escaping protocols for the telemetry strings before they are incorporated into the report template.
Boundary markers: The skill does not currently employ explicit boundary markers or "ignore instructions" warnings when processing the telemetry and Git metadata.
System Command Execution: The skill utilizes standard command-line utilities, including git and PowerShell-compatible tools like Select-String and Measure-Object. These are used appropriately within the skill's workflow to aggregate repository metrics and analyze code history.
External Resource Access: The skill connects to internal Kusto clusters and official GitHub repositories to retrieve telemetry data and report templates. These network operations are consistent with the skill's stated purpose and target the vendor's established infrastructure.

azurebackup-telemetry-report