The Agent Skills Directory

[DYNAMIC_CONTEXT_INJECTION]: The skill utilizes shell execution placeholders to automatically detect project state at load time. These commands (e.g., ls, git log, find, grep) are used solely for gathering local project metadata such as engine files, source counts, and git contributor history to streamline the onboarding workflow. These are benign, read-only operations for development purposes.
[INDIRECT_PROMPT_INJECTION]: The skill establishes a vulnerability surface by analyzing local project artifacts, including filenames and code comments (specifically grepping for 'TODO', 'FIXME', etc.) to categorize project maturity. While this involves reading local file content into the agent's context, the operations are limited to local inspection and are used to inform the provided project state report.

game-start