new-java-e2e-test-yaml-and-test

Pass

Audited by Gen Agent Trust Hub on Sep 3, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs users to run standard Java development commands using Maven, such as mvn spotless:apply and mvn failsafe:integration-test, to verify the implementation of integration tests.
  • [INDIRECT_PROMPT_INJECTION]: The skill documents an integration testing surface where the agent processes prompts and tool outputs.
  • Ingestion points: User prompt content within the test/snapshots/*.yaml files and strings passed to the session.sendAndWait method in MyFeatureIT.java.
  • Boundary markers: Prompts are delimited by structured YAML fields and Java code string literals.
  • Capability inventory: The testing framework facilitates the execution of CLI tools, such as the view tool, to simulate and verify agent interactions with the filesystem.
  • Sanitization: None, as the framework is designed to evaluate the agent's raw logic and response matching against pre-defined snapshots.
  • [SAFE]: The skill promotes the use of a mock authentication token (fake-token-for-e2e-tests) for automated testing, which is a security best practice that prevents the accidental exposure of real credentials in development or CI environments.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 3, 2026, 02:40 AM
Security Audit — agent-trust-hub — new-java-e2e-test-yaml-and-test