graphic-ebook

Pass

Audited by Gen Agent Trust Hub on Oct 11, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill's PDF export script (scripts/export-pdf.sh) installs the playwright package and the Chromium browser at runtime. These downloads target the official NPM registry and Playwright project infrastructure to enable headless browser rendering for document conversion.
  • [COMMAND_EXECUTION]: The SKILL.md instructions direct the agent to execute a local shell script (scripts/export-pdf.sh) and use standard filesystem commands (mkdir) to organize the generated ebook files. These commands are scoped to the project directory and the skill's internal assets.
  • [DYNAMIC_EXECUTION]: To handle the PDF rendering, the export-pdf.sh script dynamically generates a temporary Node.js script (export-slides.mjs) that starts a local HTTP server to serve the ebook's HTML content to the headless browser. The script is generated from a static template within the bash script.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user content (such as topic descriptions and ebook body copy) and interpolates it into the generated HTML files. While this creates a potential surface for instructions embedded in user data, the skill's output is primarily intended for rendering into static documents (HTML/PDF), limiting the agent's exposure to those instructions during the generation phase.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 11, 2026, 11:23 PM