devexpress-office-file-api-ai-powered-extensions
Pass
Audited by Gen Agent Trust Hub on Oct 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external Word, PDF, and PowerPoint documents which may contain embedded malicious instructions intended to manipulate the AI model's output or behavior.
- Ingestion points: Documents are loaded from external sources using
wordProcessor.LoadDocument,pdf.LoadDocument, and thePresentationconstructor as demonstrated inSKILL.md,references/word-processing-extensions.md,references/pdf-extensions.md, andreferences/presentation-extensions.md. - Boundary markers: The skill documentation in
SKILL.mdexplicitly mentions 'automatic prompt-injection protection' enabled by default, which adds system-level instructions to every AI request to help the model identify and disregard injected instructions. - Capability inventory: The AI-powered extensions have the capability to read and write document files and communicate with external AI providers via network requests.
- Sanitization: The skill relies on the library's built-in prompt injection protection mechanisms to sanitize or ignore instructions embedded within the document content.
Audit Metadata