extension-points

Pass

Audited by Gen Agent Trust Hub on Sep 7, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill provides standard documentation for MSBuild extension points, focusing on legitimate patterns for SDKs, NuGet packages, and repository-wide configuration. No malicious logic, obfuscation, or unauthorized access attempts were found in the skill content.
  • [INDIRECT_PROMPT_INJECTION]: The skill documents MSBuild's automatic discovery and import mechanisms (e.g., Directory.Build.props). This creates a surface where an agent analyzing an untrusted repository could be manipulated by malicious build logic embedded in those files.
  • Ingestion points: MSBuild project files and auto-discovered Directory.Build.* files in the project tree.
  • Boundary markers: Not present; MSBuild imports lack security boundaries between project and imported logic.
  • Capability inventory: MSBuild targets can execute arbitrary shell commands (Exec task), modify the filesystem, and access environment variables.
  • Sanitization: None; the skill documents standard build patterns which execute logic as-is without validation.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 7, 2026, 07:35 AM
Security Audit — agent-trust-hub — extension-points