gamma-migration-deep-dive
Pass
Audited by Gen Agent Trust Hub on Aug 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_CONTEXT_INJECTION
Full Analysis
- [DYNAMIC_CONTEXT_INJECTION]: The
SKILL.mdfile contains a dynamic command!npm list 2>/dev/null | head -10which executes when the skill is loaded to provide context about the current Node.js environment. This is used for informational purposes and does not perform sensitive operations. - [INDIRECT_PROMPT_INJECTION]: The skill facilitates the extraction of content from external presentation files (PowerPoint, Google Slides, PDFs) to be used as prompts for AI-driven generation. This creates a potential surface for indirect prompt injection if source files contain malicious instructions.
- Ingestion points: Presentation files are processed via
node:fs/promisesand the Google Drive API inSKILL.mdandreferences/implementation.md. - Boundary markers: The logic in
slidesToGammaPromptconcatenates slide text without using explicit delimiters or instructions to the model to ignore embedded commands. - Capability inventory: The skill utilizes file system access (Read/Write/Edit) and network access to communicate with the Gamma and Google APIs.
- Sanitization: The implementation extracts text directly from XML/document structures without specific sanitization or escaping of potential injection patterns within the document content.
Audit Metadata