The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to fetch and analyze information from external vendor websites (store.crowdin.com and support.crowdin.com) to suggest solutions. This creates a surface where content on those external pages could theoretically be used to influence or manipulate the agent's behavior. \n
Ingestion points: The skill fetches llms.txt indexes and individual description files from Crowdin's official store and documentation domains as directed in references/catalogs.md.\n
Boundary markers: The instructions do not define specific delimiters or instructions for the agent to treat the fetched remote content as untrusted or to ignore embedded commands.\n
Capability inventory: Although the skill itself is a triage layer, it has the capability to delegate tasks to powerful skills such as crowdin-cli, crowdin-api-client, and create-app.\n
Sanitization: The skill does not perform any validation, filtering, or sanitization of the remote text content before processing it. \n- [EXTERNAL_DOWNLOADS]: The skill fetches catalog data and help documentation from vendor-owned domains (store.crowdin.com and support.crowdin.com). These network operations are legitimate lookups used to provide accurate and up-to-date recommendations based on current platform features and available marketplace apps.

how-to-crowdin