The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions in SKILL.md and a dedicated shell script scripts/check-build.sh to execute ./gradlew build and other lifecycle commands (runClient, runServer, runData). These are standard operations for compiling and testing Java-based Minecraft mods.
[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection because it is designed to ingest and act upon data from external sources (project files and user prompts) while maintaining capabilities to modify the workspace and execute commands.
Ingestion points: The skill reads project configuration files such as gradle.properties, build.gradle, and settings.gradle to identify the modding platform and version.
Boundary markers: There are no explicit delimiters or system instructions used to separate the content of processed files from the agent's internal logic.
Capability inventory: The skill can perform file-system write operations to generate mod assets and source code, and can execute arbitrary build scripts via the Gradle wrapper.
Sanitization: Data read from external project files is utilized directly without sanitization or validation before being incorporated into the agent's reasoning or subsequent commands.

minecraft-modding