llvm
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides templates for executing standard LLVM toolchain binaries such as
clang,opt,llc, andllvm-dis. These commands are used for legitimate code compilation, optimization, and disassembly within the agent's environment. - [INDIRECT_PROMPT_INJECTION]: The skill operates on user-supplied source code and LLVM IR, which is an ingestion surface for external data.
- Ingestion points: C/C++ source code and LLVM IR files (e.g.,
src.c,src.ll) are processed by commands described inSKILL.md. - Boundary markers: No specific delimiters or instructions to ignore embedded content are defined in the command templates.
- Capability inventory: The skill utilizes shell execution for
clang,opt,llc,llvm-dis,llvm-as,llvm-link, and other diagnostic tools. - Sanitization: Input code is processed directly by the LLVM toolchain without additional sanitization layers, which is standard for compiler-related tasks.
Audit Metadata