production
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the analysis of project source code to provide testing and security recommendations, which creates a potential surface for indirect prompt injection if analyzed files contain malicious instructions.
- Ingestion points: The agent ingests local source code and project files during analysis and optimization tasks (references/GUIDE.md).
- Boundary markers: No explicit boundary markers or instructions to ignore embedded instructions are defined for the analyzed content.
- Capability inventory: The skill enables the agent to interact with the local file system and execute build commands such as gradle (SKILL.md).
- Sanitization: No sanitization or content filtering for the analyzed project data is present in the skill instructions.
- [COMMAND_EXECUTION]: The skill includes explicit instructions for the agent to execute standard build and deployment commands in the developer environment.
- Evidence: SKILL.md provides shell commands for generating production application bundles using
./gradlew bundleRelease.
Audit Metadata