The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its reliance on external data sources for code generation.
Ingestion points: The skill reads identity and styling guidelines from the brand/ directory, specifically BRAND.md, VISUAL.md, and VOICE.md in the 'Generation Flow' section.
Boundary markers: Absent. The skill does not implement delimiters or instructions to ignore potential commands embedded within the brand files.
Capability inventory: Across its templates and guides, the skill has the capability to write multiple files (Markdown, HTML, JSON, CSS), execute shell commands via package managers (npm, npx), and generate network-enabled code (CDN references).
Sanitization: The skill does not describe any validation or filtering of the content read from the brand files before interpolating it into the presentation structure.
[EXTERNAL_DOWNLOADS]: The skill references and downloads software packages and styling assets from well-known services.
Fetches the Reveal.js library and highlight plugins from the jsDelivr CDN.
Imports typography assets from the Google Fonts service.
Instructs the user to install various frameworks (Slidev, Spectacle, Decktape) via the NPM registry.
[COMMAND_EXECUTION]: The instructions and framework guides utilize shell commands for project initialization, content serving, and file transformation.
Uses npm install and npm init for dependency management.
Uses npx to execute tools like slidev, serve, and decktape for rendering and exporting presentations.
Recommends the use of system utilities like base64 for encoding assets into portable HTML files.

content-presentation