codeguard-vbnet
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted code and project metadata. Ingestion points: Processes .vb, .sln, and .vbproj files as described in the workflow. Boundary markers: Absent. Capability inventory: Can execute dotnet commands and perform file writes. Sanitization: Instructions mandate manual diff reviews and explicit user confirmation.
- [COMMAND_EXECUTION]: The skill triggers the execution of system-level developer tools. Evidence: Orchestrates dotnet format for code validation and repair. Scope: Commands are limited to pre-existing SDK tools and explicitly forbid the acquisition of new software.
Audit Metadata