dotnet-semantic-kernel
Pass
Audited by Gen Agent Trust Hub on Apr 11, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides technical documentation and templates for building AI applications. No malicious logic or unauthorized behaviors were detected.\n- [SAFE]: Credential management follows security best practices; examples use configuration abstractions and environment variables.\n- [SAFE]: Security-centric reference material explicitly details how to prevent common vulnerabilities such as plugin leakage, unauthenticated function execution, and credential hardcoding.\n- [PROMPT_INJECTION]: The skill demonstrates processing of user input and external context, representing a surface for indirect prompt injection. This is addressed via provided security mitigations in the documentation.\n
- Ingestion points: External input is processed in SKILL.md (ChatAsync) and search context is used in references/patterns.md (RagEnhancedPlugin).\n
- Boundary markers: Templates in the patterns documentation use instructional delimiters to scope the AI response.\n
- Capability inventory: The system design supports plugin-based function calling and service orchestration.\n
- Sanitization: Comprehensive guidance on input validation and result filtering is provided in the anti-patterns.md reference.
Audit Metadata