codeguard-zig
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted code files, which is a potential surface for indirect prompt injection.\n
- Ingestion points: The skill reads
.zigfiles andbuild.zigconfiguration files within the local workspace (SKILL.md).\n - Boundary markers: There are no explicit instructions for the agent to ignore prompt-like content found within the code being processed.\n
- Capability inventory: The skill utilizes the
zigcommand-line tool to perform formatting (zig fmt) and checks, which involves file system interaction.\n - Sanitization: The skill relies on the standard Zig compiler/formatter and does not perform independent sanitization of the code content.\n- [COMMAND_EXECUTION]: The skill executes local shell commands to perform its primary function.\n
- Evidence: The skill invokes
zig version,zig fmt --check ., andzig fmtto validate and modify project files. These commands are restricted to the local environment and the standard Zig toolchain.
Audit Metadata