The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's logic is driven by the content of external design documents, creating a surface where malicious input could influence agent actions.
Ingestion points: The skill instructions require the agent to read and "compress" content from GAME_DESIGN.md, ART_DIRECTION.md, and PRODUCT_BRIEF.md to create build briefs.
Boundary markers: There are no explicit delimiters or sanitization steps defined to separate design data from instructional content within these files.
Capability inventory: The agent is empowered to execute shell commands and run Python scripts based on the parameters derived from these files.
Sanitization: The skill lacks explicit sanitization or validation protocols for the text ingested from the design documents before it is used to define build commands.
[COMMAND_EXECUTION]: The skill facilitates the execution of arbitrary shell commands as part of its core build and verification workflow.
Evidence: The references/build-brief-contract.md defines a schema for commands such as install, buildOrExport, start, and verify which the agent is expected to execute.
[DYNAMIC_EXECUTION]: The skill involves the use of dynamic scripting environments to process game assets.
Evidence: references/blender-asset-workflow.md specifies the use of Blender Python and CLI for scene assembly and asset modification, which involves executing Python code within the host environment.

game-build