msbuild-antipatterns
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [Indirect Prompt Injection]: The skill processes external MSBuild files, which are untrusted ingestion points. This represents a potential surface where malicious instructions could influence agent behavior. (1) Ingestion points: Project and configuration files like .csproj, .props, and .targets (SKILL.md). (2) Boundary markers: Absent. (3) Capability inventory: File manipulation and shell execution via MSBuild tasks like and . (4) Sanitization: Absent.
- [Command Execution]: The skill documents the use of the MSBuild task for shell interaction. While intended for automation tasks like code signing and setting permissions, this involves the execution of commands within the system environment.
Audit Metadata