convert-to-cpm
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection vulnerability surface. Ingestion points: It reads untrusted repository files including project files (.csproj), solution files (.sln, .slnx), and property files (.props, *.targets) during the audit phase. Boundary markers: There are no explicit boundaries or instructions defined to isolate the untrusted project data from instructions. Capability inventory: It executes local CLI commands such as dotnet build, dotnet restore, dotnet clean, grep, and PowerShell tools. Sanitization: It does not perform validation or sanitization on the parsed project data before importing it into the context.
Audit Metadata