The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the AI agent to automatically execute a shell command (ls) upon first activation to scan several local directories (e.g., ~/.claude/skills/, ~/.agent/skills/) to check for the presence of recommended companion skills.
[EXTERNAL_DOWNLOADS]: The skill documentation encourages the installation of additional remote components using the command npx skills add AbsolutelySkilled/AbsolutelySkilled, which fetches code from an external repository.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by providing code patterns that ingest and display data from untrusted external APIs (e.g., blog post titles and excerpts) within a site's HTML metadata.
Ingestion points: Data is fetched from external sources using fetchPost in Next.js, useFetch in Nuxt, and loaders in Remix examples across the reference files.
Boundary markers: The provided patterns do not implement specific delimiters or instructions to ignore embedded instructions within the dynamic content.
Capability inventory: The skill is designed for generating HTML and metadata; while it doesn't directly expose high-risk system capabilities, the output is executed in the context of production web applications.
Sanitization: The code examples do not demonstrate sanitization or validation of the externally fetched data before it is rendered into the page structure.

on-site-seo