The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and refactor Java source code provided by the user. This establishes a vulnerability surface where malicious instructions embedded in the code could potentially influence the agent during the refactoring process.
Ingestion points: The workflow involves reading user project files in step 2 and 3.
Boundary markers: The instructions lack explicit delimiters or warnings to ignore instructions found within the user-provided code.
Capability inventory: The skill utilizes file-writing capabilities for refactoring and shell execution for Maven builds as seen in the Constraints section.
Sanitization: There is no evidence of input validation or sanitization for the code being processed.
[COMMAND_EXECUTION]: The skill mandates the use of build tools (mvn or ./mvnw) to compile and verify the project. While these are standard procedures in software development, the agent is directed to execute shell commands which depend on the user's local environment and build configuration (e.g., pom.xml).
[DYNAMIC_EXECUTION]: The requirement to run compilation and verification tests (mvn clean verify) involves the dynamic execution of code within the user's project. This is a standard but noteworthy behavior for an AI agent performing software engineering tasks, as it executes logic defined in the target project.

126-java-exception-handling