The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input in the form of asset requests (e.g., "Need a health power-up sprite") which the agent uses to perform file system searches and operations.
Ingestion points: User requests and descriptions for game assets are processed directly from the chat environment.
Boundary markers: The skill lacks explicit boundary markers or instructions to the agent to treat these user-provided strings as untrusted data when interpolating them into search patterns or command arguments.
Capability inventory: The skill utilizes the Bash tool to execute ls, find, cp, and magick (ImageMagick) commands, and the Read tool for visual asset inspection.
Sanitization: No sanitization or validation logic is specified for user input before it is utilized in shell command structures or file path assembly.
[COMMAND_EXECUTION]: The skill provides explicit instructions for the agent to use shell commands via Bash for asset discovery and manipulation. While these commands (ls, find, cp, magick) are aligned with the skill's primary purpose of asset management, they operate on the local file system using inputs derived from user requests.

game-artist