The Agent Skills Directory

[COMMAND_EXECUTION]: The skill requires the agent to execute shell commands such as ./mvnw compile, mvn compile, and ./mvnw clean verify to validate changes to the project.
[DYNAMIC_EXECUTION]: The skill uses build tools to compile and verify code that is generated or modified by the agent during the task.
[INDIRECT_PROMPT_INJECTION]: The skill processes project-specific context and external instructions to modify code and execute commands, creating a potential attack surface.
Ingestion points: Project source files including Java records, application.properties, and pom.xml, in addition to user-provided requirements.
Boundary markers: The skill does not define specific delimiters for untrusted data, although it mandates following a structured reference guide.
Capability inventory: The skill has the ability to write to the file system and execute shell commands through Maven.
Sanitization: The reference documentation includes a specific safeguard against constructing Kafka topic names or headers from untrusted user input, which acts as a mitigation for the code it generates.

414-frameworks-quarkus-kafka