utility-slideshow-creator
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill operates by executing local Node.js scripts (
scripts/generate-deck.jsandscripts/export-pdf.mjs) to process AI-generated JSON specifications. This is the primary intended function of the skill to generate presentation files. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data (a "content brief") to generate a structured JSON specification. While this presents an attack surface for indirect prompt injection, the risk is mitigated by the requirement to map content to a rigid, calibrated JSON schema before it reaches the execution phase.
- Ingestion points: User-provided content briefs (untrusted data).
- Boundary markers: No explicit delimiters for the input brief are mentioned in the instructions.
- Capability inventory: Execution of local Node.js scripts (
scripts/generate-deck.js) to generate files. - Sanitization: The instructions require the AI to translate input into a structured JSON schema with strict character limits and predefined slide types, which serves as a validation layer.
- [EXTERNAL_DOWNLOADS]: The skill uses Puppeteer for PDF export, which involves fetching resources (such as fonts) from well-known services like Google Fonts. This is documented as part of the normal operation for maintaining visual consistency.
Audit Metadata