{"code":0,"data":{"id":"d2d2ae92-d95b-4d55-b343-be411c3147d3","publisher_type":"organization","publisher_unique_handle":"langgenius","creator_email":"shani@dify.ai","template_name":"Multi-Question RAG Workflow","icon":"","icon_background":"","icon_file_key":"templates/d2d2ae92-d95b-4d55-b343-be411c3147d3/icon.png","kind":"classic","dsl_file_key":"templates/d2d2ae92-d95b-4d55-b343-be411c3147d3/dsl.yaml","dsl_raw_file_key":"","asset_files":null,"asset_tree_nodes":null,"categories":["support","knowledge"],"deps_plugins":["langgenius/cohere","langgenius/openai"],"preferred_languages":["en"],"overview":"This instructional workflow demonstrates how to build a multi‑step pipeline in Dify. It accepts a body of text, extracts all standalone questions using a parameter‑extraction node, loops through each question, retrieves relevant knowledge from a dataset, and uses an agent to draft an answer with citations. Finally, the answers are formatted into a structured reply. The template is intended for educational purposes to help new users understand loops, iterations, knowledge retrieval and agent nodes.","readme":"Prepare a knowledge base/dataset in Dify and note its dataset ID; ensure retrieval is enabled.\r\n \r\nConfigure the Parameter Extractor nodes with your preferred LLM provider (e.g., OpenAI) and the prompt to extract questions.\r\n \r\nIn the retrieval nodes, set the dataset ID and choose the desired retrieval parameters (e.g., search type and top‑k).\r\n \r\nProvide an LLM provider for the agent that generates answers and final formatting.\r\n \r\nDeploy the workflow and input text containing multiple questions to see how the pipeline extracts, searches and answers.","partner_link":"","version":"0.5.0","status":"published","review_comment":"","usage_count":825,"created_at":"2026-03-05T22:39:55.839018Z","updated_at":"2026-03-05T22:39:55.839018Z"},"msg":"ok"}