The Agent Skills Directory

[Indirect Prompt Injection Surface]: This skill presents a surface for indirect prompt injection because it reads untrusted data from container logs and telemetry streams using commands like aspire logs and aspire otel logs (see SKILL.md). The skill recommends using structured JSON formatting which provides some boundary separation, but does not explicitly instruct the agent to ignore embedded instructions within log content. The capability inventory for the agent includes executing diagnostic commands via the aspire, kubectl, and az CLI tools. No specific sanitization methods for ingested log content are described.
[Asynchronous Tool Execution]: The skill includes guidance for running background daemons like the Aspire Dashboard. it instructs agents to use asynchronous execution modes to ensure process persistence without blocking the primary workflow, which is a standard operational practice for long-running monitoring tasks.
[Context-Aware Tool Routing]: The skill defines a diagnostics bridge to ensure the agent uses environment-specific tools, such as kubectl for AKS and azure-diagnostics for Azure services. This promotes the principle of least privilege by scope-limiting tool use to the appropriate diagnostic context.

aspire-monitoring