og-image
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- [EXTERNAL_DOWNLOADS] (SAFE): The skill references standard tools like Playwright for browser-based screenshots. These are typical for web development workflows and do not involve untrusted third-party scripts.
- [COMMAND_EXECUTION] (SAFE): It uses browser automation commands (
browser_navigate,browser_resize,browser_take_screenshot) which are limited to capturing a locally generated route (/og-image) within the project environment. - [DATA_EXFILTRATION] (SAFE): No evidence of sensitive data access or exfiltration. The skill focuses on reading public branding assets (logos, colors, package name) to generate an image and saving that image to the local
/publicdirectory. - [INDIRECT_PROMPT_INJECTION] (LOW): While the skill reads untrusted data from the codebase (package.json, landing pages) to generate images and meta tags, the impact is limited to the visual content of the social image and metadata, which are subject to developer review. Mandatory Evidence:
- Ingestion points:
package.json, tailwind configs, and landing pages. - Boundary markers: None explicitly defined, but usage is restricted to static content generation.
- Capability inventory: Browser navigation, screenshotting, and file writing (local).
- Sanitization: Not explicitly present, but the scope of operation is restricted to local development files.
Audit Metadata