cpp-systems
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes C++ source files and build configurations (defined in the
pathsglob). This creates a standard indirect prompt injection surface where malicious instructions could be embedded in code comments or project metadata within a repository being reviewed. However, no active exploitation patterns were found, and the skill follows industry best practices for code analysis.\n - Ingestion points:
**/*.cpp, **/*.hpp, **/*.cc, **/*.hh, **/*.cxx, **/*.h, **/CMakeLists.txt, **/*.cmake\n - Boundary markers: None specified in the instructions.\n
- Capability inventory: Source code analysis, refactoring (file writing), and execution of local build/test tools (clang-tidy, gtest, etc.).\n
- Sanitization: Not applicable to the instructional content.\n- [COMMAND_EXECUTION]: The skill incorporates verification steps that involve running standard C++ developer tools such as
clang-tidy,clang-format, and test runners (ctest,gtest). These commands are standard for the intended domain and do not involve suspicious execution patterns or untrusted remote scripts.\n- [SAFE]: The skill provides high-quality technical documentation and verification steps using well-known, legitimate developer tools. It explicitly recommends against including sensitive data like credentials or machine-specific paths in the repository, aligning with security best practices.
Audit Metadata