The Agent Skills Directory

[COMMAND_EXECUTION]: The validation script assets/co-op-lab/tools/check_coop.py uses subprocess.run to invoke the system C compiler and execute generated test binaries. The command construction uses fixed paths and strictly validated internal variables, preventing shell injection.
[DYNAMIC_EXECUTION]: The skill performs runtime compilation of its C source code (specifically coop.c and input.c) in assets/co-op-lab/tools/check_coop.py to facilitate unit testing on the host machine before deploying to the target platform.
[EXTERNAL_DOWNLOADS]: The build instructions in assets/co-op-lab/README.md utilize the ewrogers/megadev Docker image. This is a resource owned by the skill author and serves as the official build environment for the project.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied project metadata during setup. However, it implements strong security controls:
Ingestion points: Project id and title arguments in scripts/create_project.py.
Boundary markers: None (CLI arguments).
Capability inventory: File system writing via shutil.copytree and Path.write_text in scripts/create_project.py.
Sanitization: Strict regex validation (r'[a-z][a-z0-9_]{0,10}' for ID and r'[A-Za-z0-9][A-Za-z0-9 _-]{0,47}' for title) effectively prevents directory traversal or command injection attacks.

megadev-multiplayer