skills/soujava/agent-skills/sdd4j/Gen Agent Trust Hub

sdd4j

Pass

Audited by Gen Agent Trust Hub on Oct 7, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill invokes build and test commands (such as ./mvnw test) to verify that the implementation conforms to the capability specifications. These commands are detected from the project environment during the setup phase and are standard for Java development workflows.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes natural language inputs, including feature descriptions and README content, to derive specifications and code structures. This represents a potential surface for indirect injection if project documentation contains malicious instructions.
  • Ingestion points: The new mode reads feature descriptions and README.md seeds to bootstrap specifications.
  • Boundary markers: The skill instructions mandate a clarification loop to resolve ambiguities and require explicit user confirmation before writing or modifying files.
  • Capability inventory: The skill is capable of writing Java source code, generating tests, and executing system commands via the build tool.
  • Sanitization: Security is maintained through human-in-the-loop review; the agent is instructed to ask for confirmation before committing changes based on its interpretation of the documentation.
  • [DYNAMIC_EXECUTION]: The skill generates Java implementation and test code at runtime based on the capability specifications. This code is then compiled and executed by the project's existing infrastructure to perform verification. This behavior is the primary intended purpose of the skill and is constrained by the project's architecture and stack settings.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 7, 2026, 01:15 PM