recoup-internal-consulting-graphics
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes
npx playwright screenshotto render generated HTML files into PNG images. This is the primary method used to produce the requested graphics. - [EXTERNAL_DOWNLOADS]: Fetches static assets, including fonts and branding logos, from Vercel's public blob storage (
public.blob.vercel-storage.com). These references target a well-known service and are necessary for the visual rendering process. - [INDIRECT_PROMPT_INJECTION]: The skill ingests user content to populate placeholders in HTML templates (e.g.,
{{WHAT_IT_MEASURES}},{{FIGURE}}). While no explicit sanitization is performed, this is inherent to its function as a template-based graphic generator. - Ingestion points: Content provided by the user for graphics in
SKILL.md. - Boundary markers: Not explicitly implemented in the HTML template files.
- Capability inventory: Shell execution (Playwright rendering) and local file system writes (via the internal
materialize.mjsscript). - Sanitization: No explicit escaping or filtering of user content is defined before it is placed into HTML shells.
Audit Metadata