devexpress-office-file-api-ai-powered-extensions
Pass
Audited by Gen Agent Trust Hub on Aug 18, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill documents tools that ingest untrusted document data (Word, PDF, PowerPoint) and send it to AI models, creating a potential indirect prompt injection surface. The documentation explicitly addresses this and describes built-in mitigations.
- Ingestion points: External document content is loaded via
wordProcessor.LoadDocument,pdf.LoadDocument, and thePresentationconstructor across all reference files and examples. - Boundary markers: The skill notes that the library includes system-level instructions by default to help the LLM identify and disregard injected instructions.
- Capability inventory: The service possesses capabilities to read and write document files to the local file system and perform network requests to AI providers (e.g., Azure OpenAI, OpenAI, Google Gemini).
- Sanitization: The documented library claims to provide automatic protection against malicious content that attempts to override system behavior.
- [EXTERNAL_DOWNLOADS]: The skill provides standard instructions to install NuGet packages using the .NET CLI. All identified packages are from trusted organizations (Microsoft, OpenAI) or are vendor-specific resources (DevExpress).
- [CREDENTIALS_SAFE]: The code examples use descriptive placeholders like
YOUR_AZURE_OPENAI_KEYand explicitly advise the use of environment variables or secrets managers rather than hardcoding credentials in production code.
Audit Metadata