spring-boot-service-scaffold

Pass

Audited by Gen Agent Trust Hub on Jun 17, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill performs mandatory build and test verification using shell commands mvn -q -DskipTests verify and mvn -q test. This executes the agent-generated project code in the local environment to ensure quality.\n- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) because it ingests architectural directions and service metadata to generate executable build configurations. Malicious input could cause the generation of a pom.xml or build.gradle.kts file that executes harmful code during the verification phase.\n
  • Ingestion points: SKILL.md (Step 1) ingests service name, organization name, and architectural directions.\n
  • Boundary markers: There are no explicit instructions to wrap or delimit untrusted architectural inputs to prevent instruction confusion during generation.\n
  • Capability inventory: The skill executes shell commands via mvn for verification.\n
  • Sanitization: The skill sanitizes the Java package root by stripping hyphens but does not validate the logic within the architectural requirements.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 17, 2026, 08:35 AM
Security Audit — agent-trust-hub — spring-boot-service-scaffold