The Agent Skills Directory

Command Execution: The skill utilizes the Bash tool to execute ms CLI commands for application management. These operations are core to the skill's purpose and are conducted using established vendor tooling.
Local File Access: The skill reads configuration from ms.config.json and memory-bank.md to identify the target application. This allows the agent to maintain context within the local development environment.
Safe Confirmation Patterns: A significant safety feature is the requirement for two separate manual confirmations from the user before the application is deleted from the service. The skill specifically avoids automatically adding the --force flag, ensuring the user is informed if the service detects child resources or active shares.
Indirect Prompt Injection Surface: As the skill processes data from local files like ms.config.json (ingestion point) to execute commands via Bash (capability), there is a theoretical surface for indirect injection if these files were maliciously modified. However, the requirement for manual user confirmation (sanitization) effectively mitigates the risk of the agent performing unintended actions based on that data.
Instructional Safety: The skill provides the user with local cleanup commands as suggestions rather than executing them automatically, maintaining user control over the local file system.

delete-app