rust-unsafe-ffi
Pass
Audited by Gen Agent Trust Hub on Jul 31, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill contains non-ASCII characters ('爱国') within code snippets in 'references/ffi-bindgen.md' (e.g., '.bindgen爱国'). This is invalid Rust syntax and represents suspicious content that could be used as a 'canary' or trigger token for specialized agent behaviors or to influence model output in unexpected ways.
- [PROMPT_INJECTION]: The skill documents a surface for indirect prompt injection through the ingestion of untrusted external data.
- Ingestion points: The 'SKILL.md' and 'references/ffi-bindgen.md' files provide instructions for using the 'bindgen' tool to parse external C/C++ header files, such as 'wrapper.h' and 'cpp_wrapper.hpp'.
- Boundary markers: Absent. The skill does not include instructions or delimiters to warn the agent about potentially malicious instructions embedded in the source files being parsed.
- Capability inventory: The skill involves the use of 'cargo build' and the 'bindgen' library, which processes external source code to generate Rust bindings.
- Sanitization: Absent. The provided examples do not demonstrate validation or sanitization of the header file content before it is processed by the build system.
Audit Metadata