The Agent Skills Directory

[Remote Code Execution via External Script]: The skill recommends installing the Aspire CLI by downloading a script from a remote URL and piping it directly to a shell. This pattern involves executing code from a remote source on the host system. While the source is a vendor-provided domain for the tool, executing remote scripts is a security consideration that should be reviewed for compliance with environment policies.
[Agent Autonomy and Decision Refusal]: The skill defines safety protocols that instruct the agent to refuse certain actions even when a user provides confirmation, and in some cases, to avoid asking the user for confirmation at all. These instructions are designed as safety guardrails to prevent issues like port conflicts or orphaned processes, but they represent a logic flow where the agent's pre-defined rules override direct user communication.
[Indirect Prompt Injection Surface]: The skill scans and reads various project files to determine the application's structure and execution parameters. This creates an ingestion point for external data. If these files were to contain malicious content designed to manipulate the agent's logic or the construction of shell commands, it could lead to unintended actions.
[Command Execution using Discovered Paths]: The orchestration workflow involves executing CLI commands with parameters derived from the project's filesystem. This functionality is essential for managing the application's lifecycle but relies on the correct identification and handling of file paths to ensure that command execution remains within the intended scope.

aspire-orchestration