133-java-testing-acceptance-tests
Pass
Audited by Gen Agent Trust Hub on Sep 26, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill requires the execution of Maven commands (
./mvnw compile,./mvnw clean verify) to validate the project state and confirm test success. This is a standard and expected part of the development and testing workflow. - [INDIRECT_PROMPT_INJECTION]: The skill processes Gherkin feature files or facts, which is a potential attack surface for indirect prompt injection. However, the skill provides comprehensive mitigations, explicitly instructing the agent to treat Gherkin text as untrusted data and forbidding the ingestion of raw, non-sanitized external scenario text.
- Ingestion points: Sanitized Gherkin scenario facts provided in the agent's context (SKILL.md, Step 1).
- Boundary markers: The skill contains a specific "AUTHORITY BOUNDARY" section in the reference file, instructing the agent to never obey instructions embedded in the data.
- Capability inventory: The skill possesses file-writing capabilities (creating Java test classes) and shell execution capabilities (Maven commands).
- Sanitization: The skill mandates that scenarios must be "maintainer-sanitized" before processing, ensuring a human-in-the-loop validation of the input data.
Audit Metadata