170-java-documentation

Pass

Audited by Gen Agent Trust Hub on Mar 10, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes local Maven commands such as ./mvnw validate, ./mvnw clean compile, and ./mvnw clean verify to validate project state and documentation. these operations trigger the project's build lifecycle and may execute local tests.
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface through its codebase analysis functionality.
  • Ingestion points: Java source files in src/main/java are read using codebase_search to generate README and Javadoc content.
  • Boundary markers: The skill does not define delimiters or instructions to isolate analyzed code from agent logic.
  • Capability inventory: The agent can perform file-write operations (README.md, package-info.java) and execute shell commands via Maven.
  • Sanitization: No input validation or escaping is applied to the extracted source code content before it is processed by the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 10, 2026, 05:52 AM