The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill implements a Retrieval-Augmented Generation (RAG) architecture that processes external data to provide context for AI responses.
Ingestion points: Documents and external URLs are ingested through the /ingest and /ingest/url endpoints in templates/document-ingestion.ts and templates/rag-chat.ts.
Boundary markers: templates/rag-chat.ts employs structural separators and title headers to delimit retrieved context within the LLM prompt.
Capability inventory: The skill utilizes fetch for network operations, Cloudflare Workers AI for model execution, and Vectorize for database operations.
Sanitization: A robust text extraction utility in templates/document-ingestion.ts uses HTMLRewriter to safely strip scripts and other dangerous elements from ingested HTML content.
[EXTERNAL_DOWNLOADS]: The skill integrates with trusted services for AI capabilities.
Utilizes the official OpenAI API and Cloudflare's Workers AI platform for generating embeddings.
Provides functionality to fetch content from remote URLs for the intended purpose of document ingestion and indexing.
[COMMAND_EXECUTION]: The documentation includes standard management commands for the Cloudflare platform.
Instructs the user on utilizing wrangler CLI for index creation, metadata indexing, and vector management, which are standard development procedures.

cloudflare-vectorize