315-frameworks-spring-mongodb
Pass
Audited by Gen Agent Trust Hub on Sep 26, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructions mandate the execution of shell commands through the Maven wrapper (./mvnw compile and ./mvnw clean verify). These actions are contextually appropriate for a software engineering task to ensure code integrity and functional correctness before and after applying changes.
- [INDIRECT_PROMPT_INJECTION]: The skill operates on project source code and configuration files, which are external inputs to the agent. While this creates a theoretical surface for injection, the skill includes robust mitigation instructions. 1. Ingestion points: Project files, including pom.xml and Java document/repository classes, are read by the agent to assess context. 2. Boundary markers: Explicit instructions require the agent to prioritize the reference documentation and stop immediately if build failures occur. 3. Capability inventory: Access to the local shell for running Maven build and test cycles. 4. Sanitization: The skill mandates the use of the Spring Data Criteria API and derived finders to prevent query-based injection attacks, explicitly forbidding string concatenation of user input.
Audit Metadata