single-slide-ppt
Fail
Audited by Gen Agent Trust Hub on Feb 28, 2026
Risk Level: HIGHCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides code templates in
SKILL.mdandIMAGE_GUIDE.mdthat usechild_process.execSyncto run shell commands. - [COMMAND_EXECUTION]: The
downloadImagefunction template uses unsanitized variables (url,outputPath) within a shell command:execSync(curl -o ${outputPath} "${url}", ...). This creates a command injection risk if the URL or file path is derived from untrusted input and contains shell metacharacters. - [EXTERNAL_DOWNLOADS]: The skill fetches architecture diagrams and product assets from the official Visual Studio Code documentation to enhance slides.
- [PROMPT_INJECTION]: The skill features a surface for indirect prompt injection (Category 8). It is instructed to fetch content from the web to extract URLs for download and inclusion in slides.
- Ingestion points: Web content retrieved via
fetch_webfor URL extraction. - Boundary markers: No explicit delimiters are used to separate fetched content from instructions.
- Capability inventory: Use of
child_process.execSyncfor command execution andpptx.writeFilefor file system operations. - Sanitization: The provided templates lack sanitization for URLs retrieved from external sources before they are passed to the shell.
Recommendations
- AI detected serious security threats
Audit Metadata