frontend-slides
Pass
Audited by Gen Agent Trust Hub on Apr 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes platform-specific commands to open generated presentation files, including
openfor macOS,xdg-openfor Linux, andstartfor Windows. It also callspython3to execute conversion scripts. - [EXTERNAL_DOWNLOADS]: The skill fetches typography assets from well-known services like Google Fonts and Fontshare. Additionally, it may prompt for the installation of the
python-pptxlibrary if it is not present in the execution environment. - [PROMPT_INJECTION]: The skill has an indirect prompt injection surface due to the processing of untrusted content into executable code.
- Ingestion points: External data is ingested from user notes, text drafts, and local
.pptor.pptxfiles during the conversion process. - Boundary markers: The instructions do not include the use of delimiters or specific guidance to ignore embedded instructions within the source data.
- Capability inventory: The skill performs file system write operations for HTML presentations and preview files, executes shell commands for opening files, and runs Python code for data extraction.
- Sanitization: There are no explicit requirements for sanitizing or escaping the text and assets extracted from source documents before they are embedded into the generated HTML and JavaScript code.
Audit Metadata