The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill is designed to invoke C# methods at runtime using reflection. It allows for the instantiation of types and the calling of any method, which bypasses standard object-oriented encapsulation and access modifiers.
[REMOTE_CODE_EXECUTION]: By providing a mechanism to call any function within the project codebase, the skill serves as a high-impact execution vector. An attacker who can influence the agent's input can use this to execute arbitrary logic in the context of the running application.
[PRIVILEGE_ESCALATION]: The skill explicitly advertises the ability to call private methods. This allows for unauthorized interaction with internal application logic and data that is not intended to be exposed to external interfaces.
[INDIRECT_PROMPT_INJECTION]: The skill is highly susceptible to indirect injection attacks.
Ingestion points: The filter, targetObject, and inputParameters fields in SKILL.md accept complex JSON representing method calls.
Boundary markers: None are present; there are no instructions for the agent to distinguish between valid developer intent and instructions hidden in untrusted data.
Capability inventory: Full access to the C# reflection API, enabling method invocation and object creation.
Sanitization: No sanitization or validation logic is defined to restrict which methods can be called or to verify the safety of the parameters provided.
[COMMAND_EXECUTION]: The skill utilizes a command-line interface tool, unity-mcp-cli, to bridge the agent's instructions to the target execution environment, facilitating direct system-level interaction.

reflection-method-call