The Agent Skills Directory

[COMMAND_EXECUTION]: The documentation in SKILL.md and references/blog-figure-kit.md instructs the user to execute a local script named build.py for site generation and deployment. Recommending the execution of local scripts is a known risk factor if the script's behavior is not audited, although it is common in static site generation workflows.
[EXTERNAL_DOWNLOADS]: The skill references assets from well-known technology providers. Specifically, assets/gallery.html fetches CSS and typography from Google Fonts (fonts.googleapis.com), and references/blog-figure-kit.md provides instructions for embedding and capturing content from Twitter/X via its official domain (platform.twitter.com).
[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to process user-supplied 'claims' (natural language sentences) to generate HTML and SVG visualizations. This ingestion surface is susceptible to indirect prompt injection attacks where a malicious claim could attempt to manipulate the generated output's structure or styles.
Ingestion points: User-provided declarative sentences in the 'Method' section of SKILL.md.
Boundary markers: None explicitly implemented; the input is intended to be rendered directly into a <figcaption> element.
Capability inventory: Generation of SVG code, CSS styles, and HTML structure across all reference files.
Sanitization: No specific sanitization or escaping rules are defined for the input 'claims' before they are interpolated into the generated code.

editorial-illustrations