The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from user-provided HTML templates and source content files which could potentially contain malicious instructions meant to influence the agent's behavior.
Ingestion points: Custom HTML files located in the references/templates/ directory and user-provided reference files (e.g., Markdown, Text, Word) ingested during the content hearing phase.
Boundary markers: The instructions do not define specific delimiters or "ignore instructions" warnings to wrap this external content when it is read into the agent's context.
Capability inventory: The skill has the capability to write multiple HTML files to the local filesystem and to invoke the /pptx skill for file conversion.
Sanitization: There is no explicit requirement for the agent to sanitize or filter the content of these external files before processing them.
[EXTERNAL_DOWNLOADS]: Fetches CSS frameworks and JavaScript libraries including Tailwind CSS, Font Awesome, and Chart.js from jsDelivr, a well-known and established Content Delivery Network (CDN).
[REMOTE_CODE_EXECUTION]: Instructs the user on how to install an additional /pptx skill from a trusted organization's repository on GitHub (anthropics/claude-code-agent-skills).
[DYNAMIC_EXECUTION]: Generates HTML slide files that incorporate JavaScript for data visualization (Chart.js) and a self-contained slide viewer (index.html) with interactive navigation logic. This is a primary function of the skill and is implemented using standard, transparent techniques.

slidekit-create