The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes local shell commands including base64, tr, and rm to process search query data, and uses the unity CLI to communicate with the Unity Editor.
[DYNAMIC_EXECUTION]: The skill generates C# code at runtime based on a static template to interact with the Unity Editor's SearchService. This code is executed via the unity command eval interface. This is a controlled execution of a specific UI-related API (SearchService.ShowWindow).
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest natural language input from users to generate search queries, creating a potential surface for indirect injection.
Ingestion points: User-provided search requests and asset/scene object names processed in SKILL.md.
Boundary markers: None explicitly defined for user input; however, the skill provides strict translation rules in references/query-patterns.md.
Capability inventory: Local shell execution and arbitrary C# execution via unity command eval as described in references/open-search-window.md.
Sanitization: The skill implements a critical mitigation by requiring the agent to Base64-encode the generated search query before interpolating it into the C# snippet. This prevents user-controlled text from escaping C# string literals or shell command quoting, ensuring the query is treated as data rather than code.

generate-editor-search-query