The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates the creation of Retrieval-Augmented Generation (RAG) workflows that ingest untrusted external data into the LLM context.
Ingestion points: The skill utilizes PyPDFLoader("document.pdf") and SimpleDirectoryReader("./docs") to load data from the local file system into the agent's prompt context.
Boundary markers: The provided code snippets demonstrate direct interpolation of retrieved text into messages (e.g., content: f"Context:\n{context}\n\nQuestion: ...") without using delimiters or instructions for the model to ignore embedded commands.
Capability inventory: The skill possesses the capability to read files, write to a local vector database (chromadb.PersistentClient), and perform network operations via LLM API calls.
Sanitization: The instructions do not include methods for sanitizing, filtering, or validating the external content before it is processed by the AI.
[COMMAND_EXECUTION]: The skill documentation includes shell commands for interacting with the Ollama CLI, specifically ollama signin, ollama run, and ollama pull. These commands involve user-initiated process execution and authentication with an external service.

ollama-rag