rag-local — OKF Bundle for MCP Integration
DraftLocal-first RAG memory for LLMs: Node server with ONNX embeddings, OKF-structured chunks, portable vector bundles, and REST + Web UI + CLI + MCP interfaces
Best for
- AI agents using the MCP protocol
- RAG (Retrieval Augmented Generation) pipelines
What it does
- Exposes functionality through the Model Context Protocol (MCP) for agent-native access
- Portable knowledge package that AI agents can load and use at runtime
- Version-controlled with standard Git workflows for collaboration
How to use with AI agents
Step 1: Clone the repository
git clone https://github.com/MauricioPerera/rag-local
cd rag-localStep 2: Point your AI agent at the bundle directory
# Claude Code
claude /path/to/rag-local# Cursor
cursor /path/to/rag-local# OpenCode
opencode /path/to/rag-localMCP: This bundle supports the Model Context Protocol. Configure your MCP client to discover and load this bundle.
Frequently Asked Questions
Is rag-local OKF conformant?
No, rag-local is currently marked as a draft bundle. It may not fully follow the OKF specification yet.
How do I install rag-local?
Clone the repository and point your AI agent at the bundle directory. The agent will automatically discover and load the OKF knowledge. No additional tools or SDKs are required.
What AI agents can use rag-local?
rag-local works with any AI agent that supports OKF bundles, including Claude Code, Cursor, OpenCode. Just point your agent at the bundle directory.
What are alternatives to rag-local?
Similar bundles include crystalline, mnemos, agentic-task-system. Browse these for alternative approaches and feature sets.