The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes structured content data (titles, narrative, stats) and interpolates it directly into HTML templates. If the input data contains malicious instructions or scripts, they could influence agent behavior or be executed during the rendering process.
Ingestion points: content data input field and Asset Types configuration in SKILL.md.
Capability inventory: File writing to the local filesystem and shell command execution using npx playwright for PDF generation.
Sanitization: The instructions do not specify any escaping or sanitization protocols for the data injected into the HTML templates.
Boundary markers: There are no explicit markers or delimiters defined to isolate untrusted content from the template instructions.
[EXTERNAL_DOWNLOADS]: The skill interacts with external APIs including Gamma (api.gamma.app) and Vercel's v0 platform (api.v0.dev). It also retrieves resources from well-known services such as Google Fonts and the Tailwind CSS CDN. These operations are consistent with the skill's purpose of asset creation.
[COMMAND_EXECUTION]: The skill uses shell commands via npx playwright to convert generated HTML files into PDFs. This is an expected functional component of the document creation workflow.
[DYNAMIC_EXECUTION]: The skill dynamically generates HTML files using templates and injected content, which are then rendered by a headless browser (Playwright). This pattern represents runtime execution of generated script content.

content-asset-creator