The Agent Skills Directory

[COMMAND_EXECUTION]: The build script (assets/combat-lab/tools/build.py) executes docker to compile the C source into a ROM. The command is safely constructed as a list, which prevents command injection vulnerabilities by avoiding shell-based argument parsing.
[EXTERNAL_DOWNLOADS]: The build process fetches a specialized SDK environment from the author's Docker Hub repository (ewrogers/sgdk). This is a legitimate vendor resource used for the project toolchain.
[DYNAMIC_EXECUTION]: Python scripts in the tools directory procedurally generate art and audio assets at build time. This includes generating C header files containing font and UI data, which are then compiled into the final binary.
[INDIRECT_PROMPT_INJECTION]: The skill provides instructions for the agent to integrate combat logic into user-provided project codebases.
Ingestion points: User-provided game logic and geometric collision queries in the SGDK project (SKILL.md).
Boundary markers: None present; the skill operates under the assumption of a trusted developer environment.
Capability inventory: The skill includes file system write operations (prepare.py) and Docker-based command execution (build.py).
Sanitization: The build scripts utilize path resolution and strict argument validation through argparse to mitigate risks of path traversal or command manipulation.

sgdk-combat