The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided content (titles, outlines, and brand names) to populate HTML templates, which constitutes an attack surface for indirect prompt injection.
Ingestion points: User-supplied fields for 'title', 'subtitle', 'brand', and 'sections' are processed in generate-garden-page.js (e.g., in the renderPage and renderSection functions).
Boundary markers: There are no explicit delimiters or system instructions to ignore embedded commands within the processed data.
Capability inventory: The skill has access to the Bash and Write tools, which are used to manage files and execute the Playwright automation.
Sanitization: User-provided strings are interpolated directly into the HTML template without escaping, which could allow for the injection of malicious HTML or scripts into the visual output.
[COMMAND_EXECUTION]: The skill uses the Bash tool to perform legitimate setup and execution tasks, such as installing browser binaries and running the screenshot automation script via Playwright.
[EXTERNAL_DOWNLOADS]: The skill initiates downloads of browser binaries using the official Playwright CLI tool (npx playwright install chromium). This involves downloading resources from well-known infrastructure managed by Microsoft for the purpose of enabling the skill's primary functionality.

infographic-generator