Build a local vector knowledge base with semantic search and reranking.
npx skills add https://github.com/nigo81/nigo-skills --skill local-rag-nigo81
Please help me install this Agent Skill. Skill: local-rag Source: https://github.com/nigo81/nigo-skills/tree/main/local-rag Command: npx skills add https://github.com/nigo81/nigo-skills --skill local-rag-nigo81
This Skill provides a powerful local vector knowledge base for efficient documentation management and retrieval, streamlining tasks like document ingestion, semantic search, and reranking.
Run the command local-rag search my-project "审计法规" --rerank to search for "审计法规" within your local knowledge base and get reranked results.
High-intent search queries and answers about installing and using this skill.
Build a local vector knowledge base by ingesting multiple document formats into ChromaDB, enabling semantic search and reranking. This allows efficient documentation retrieval and management directly on your local machine.
Yes, embeddings and reranking can use Ollama or OpenAI as alternative providers. The system defaults to the SiliconFlow API but supports configuring other embedding and reranker models for your local vector knowledge base.
The local RAG system supports ingestion of multiple document formats. It utilizes python-docx for Word documents and PyMuPDF for PDF files, allowing you to populate your local vector knowledge base with diverse documentation.
Semantic search reranking improves audit document retrieval by refining search results based on relevance scores. This ensures auditors efficiently retrieve precise regulations or guidelines, maintaining compliance and accuracy during complex projects.
Yes, ChromaDB is required for local execution as it provides the vector database foundation. You also need python-docx and PyMuPDF installed to handle document parsing and ingestion for the semantic search functionality.
The best way is using the semantic search command with the rerank flag, such as `local-rag search my-project "query" --rerank`. This queries your local vector knowledge base and returns refined, relevant regulation results.
QredenceIngest markdown files into ChromaDB for semantic procedural memory retrieval.
overviewlabsSearch local Markdown collections with hybrid keyword and vector retrieval.
zerwizStores embeddings and metadata in a vector database for semantic search and retrieval.
adeboldPerform vector-based semantic search against an AgentDB index.
