Load FAISS indexes and retrieve content via dot-path queries.
npx skills add https://github.com/sjtu-sai-agents/EvoMaster --skill rag-sjtu-sai-agents
Please help me install this Agent Skill. Skill: rag Source: https://github.com/sjtu-sai-agents/EvoMaster/tree/main/evomaster/skills/rag Command: npx skills add https://github.com/sjtu-sai-agents/EvoMaster --skill rag-sjtu-sai-agents
Rag enables knowledge-grounded AI by retrieving relevant information from local FAISS indexes and external knowledge sources to ground reasoning and improve accuracy.
Use the rag skill to perform a semantic search against a local vector store by pointing vec_dir to evomaster/skills/rag/MLE_DATABASE/node_vectorstore/draft (or improve) and providing a query like 'What is the main topic?' to retrieve relevant content and knowledge.
High-intent search queries and answers about installing and using this skill.
Semantic search against a local FAISS vector store is performed by pointing the skill to your vector directory and providing a query. It uses transformer embeddings to retrieve relevant content and related metadata.
Retrieval-augmented generation grounds reasoning by loading FAISS indexes and optional nodes_data to fetch relevant external documents. This contextual knowledge is exposed via dot-path queries to improve accuracy.
Yes, transformer embeddings are required to build a document Q&A assistant. The skill uses them alongside FAISS vector stores to map queries to indexed documents and extract knowledge-grounded answers.
Yes, you can extract content from task data using dot-path queries. The skill exposes retrieved knowledge and metadata flexibly, allowing targeted access to specific information within the loaded nodes_data.
The best way to retrieve knowledge from indexed documents is using FAISS and transformer embeddings for semantic search. This approach quickly fetches relevant content by matching query intent with vectorized document data.
FAISS for knowledge-grounded tasks requires a pre-built local vector index and depends on CPU-based processing. It does not natively generate answers but retrieves relevant context for downstream reasoning.
vecearBuild document embedding and retrieval pipelines for scalable semantic search.
JhabbigBuild retrieval-augmented generation systems that ground LLM answers in external knowledge.
qcmuuGenerate semantic embeddings from text using sentence-transformers models.
rizaldiemBuild retrieval-augmented generation pipelines with vector databases and document chunking.
