The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of console logs, errors, and stack traces into the agent's context, which represents a potential attack surface for indirect prompt injection if the application being debugged logs content influenced by external users.
Ingestion points: Skills such as debug_get_logs, debug_get_errors, debug_get_stack_trace, and the aggregated unity_diagnose retrieve and present console output to the agent.
Boundary markers: The instructions lack explicit delimiters or guidance to the agent for distinguishing logged data from system instructions.
Capability inventory: The skill includes high-impact capabilities such as modifying project defines and forcing domain reloads, which could be misused if an injection is successful.
Sanitization: There is no mention of sanitization, filtering, or escaping for the log data being processed.
[DYNAMIC_EXECUTION]: The skill provides mechanisms to trigger the compilation of source code and the reloading of the application domain within the Unity environment.
Evidence: The debug_force_recompile skill specifically requests assembly compilation and a domain reload. The debug_set_defines skill modifies scripting defines in PlayerSettings, which also triggers a project-wide recompilation. While these are intended features for debugging, they involve the runtime generation and loading of executable code.

unity-debug