The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill implements patterns for Retrieval-Augmented Generation (RAG), which involves ingesting external data that can be used to influence the agent's behavior.
Ingestion points: Data is loaded from the filesystem (./docs) via DirectoryLoader as seen in the Python code samples in SKILL.md.
Boundary markers: The provided prompt templates include structural delimiters like Context: and Question: to isolate retrieved content, which helps prevent accidental instruction following but is not a complete defense against adversarial content.
Capability inventory: The skill demonstrates capabilities for reading local files, communicating with external vector databases (Pinecone, Weaviate), and interacting with LLM APIs.
Sanitization: The code samples do not include explicit data sanitization or content filtering logic for the ingested documents, relying instead on LLM-based prompting strategies.

rag-implementation