Smoothie – compile a folder of data into a cited artifact agents can query
Smoothie is a tool that compiles a folder of data files into a single cited artifact, such as a PDF or markdown document, which LLM agents can then query to retrieve specific information.
Background
- Smoothie is an open-source tool (by GitHub user 4tyone) that takes a folder of unstructured data—PDFs, text files, images, audio—and compiles it into a "cited artifact" that LLM agents can query. The output is a single file that bundles the source data with a searchable index, letting AI models retrieve specific facts along with their original citations.
- The project addresses a growing problem in enterprise AI: RAG (retrieval-augmented generation) pipelines typically require running a vector database, embedding models, and an orchestrator, which is heavy and brittle. Smoothie collapses that into one portable step.
- It fits into the "context engineering" space—tools that prepare ground truth for LLMs so they don't hallucinate or miss specifics. Similar to projects like yaml-rag or llamaindex's ingestion pipeline, but aimed at maximum simplicity: compile once, query anywhere.
- The name is a playful contrast to "Blender" (a well-known 3D tool); this "smoothie" blends data instead of geometry.