322-frameworks-spring-boot-testing-integration-tests
Pass
Audited by Gen Agent Trust Hub on Sep 26, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructions require the agent to execute build and verification commands including ./mvnw compile and mvn clean verify. While standard for Java development, this represents a shell command execution surface.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to audit and improve Java source code and Maven configuration files, which serve as untrusted ingestion points. The combination of processing these files and having the capability to run build commands creates an indirect prompt injection surface.
- Ingestion points: Java test classes, pom.xml, and project structure are read and processed by the agent.
- Boundary markers: The instructions do not specify any markers or rules to ignore instructions potentially embedded within the project files being audited.
- Capability inventory: The skill has the capability to execute shell commands for project compilation and verification (mvn, ./mvnw).
- Sanitization: There is no requirement or logic for sanitizing or validating the contents of the audited project files before they are involved in the build process.
Audit Metadata