vanilla-rag · service and data flow

Dense vector retrieval with one bounded context window and one generation call

vanilla-rag service and data flow Corpus ingestion into Weaviate and a query flow through LiteLLM, the RAG plugin, embedding, dense retrieval, and generation. ATLAS COMPOSE STACK · RAG TRACK CORPUS PREPARATION · ONCE PER INGESTION QUERY PATH · ONCE PER REQUEST Corpus documentsselected dataset/profile Atlas ingestionparse · chunk · metadata Embedding modelnomic-embed-text Weaviate · RagBasechunk text + vectors CallerOpen WebUI / API LiteLLMmodel: vanilla-rag Vanilla pluginretrieve · prompt · normalize Query embeddingnomic-embed-text Weaviate · RagBasedense cosine search light_genLiteLLM → configured LLM OpenAI responseanswer · sources · metrics 1chat request 2alias dispatch 3question 4nearVector · k=5 5ranked chunks 6answer + evidence MESSAGE CONTRACT AND EXECUTION PHASES 0 · INDEXchunks → embedding vectors → RagBaseNo query-time graph or enrichment.Index is reused across requests. 1–3 · DISPATCH + EMBEDPOST /v1/chat/completions · model=vanilla-ragPlugin embeds only the incoming question.Default model properties remain alias-scoped. 4–5 · RETRIEVEnearVector(question_vector, limit=5)Top chunks retain source and dataset metadata.No BM25 fusion and no reranker. 6 · GENERATE + RETURNquestion + bounded retrieved context → light_genPlugin emits OpenAI-compatible answer payload.Evidence and latency metrics are preserved.