HaseebKhalid1507/VelociRAG
Lightning-fast RAG for AI agents. ONNX-powered, 4-layer fusion, MCP server. No PyTorch.
Combines four independent retrieval methods—vector embeddings (FAISS), full-text search (SQLite FTS5), knowledge graph traversal, and metadata filtering—fused via reciprocal rank fusion with cross-encoder reranking, all running on ONNX Runtime without PyTorch. Provides MCP server integration for Claude/Cursor/Windsurf agents, a Unix socket daemon maintaining warm model state for sub-10ms searches, and incremental graph updates that detect file changes and only rebuild affected nodes. Targets developers building AI agents that need fast, multi-modal retrieval without external APIs or GPU dependencies.
3 stars and 1,078 monthly downloads. Available on PyPI.
Stars
3
Forks
—
Language
Python
License
MIT
Category
Last pushed
Mar 28, 2026
Monthly downloads
1,078
Commits (30d)
0
Dependencies
10
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/mcp/HaseebKhalid1507/VelociRAG"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
shinpr/mcp-local-rag
Local-first RAG server for developers using MCP. Semantic + keyword search for code and...
nkapila6/mcp-local-rag
"primitive" RAG-like web search model context protocol (MCP) server that runs locally. ✨ no APIs ✨
AmberLee2427/nancy-brain
Nancy's RAG backend and HTTP API/MCP server connectors.
tac0de/knowledge-to-action-mcp
MCP server for Obsidian GraphRAG, agent-ready context, preview-only planning, and safe repo handoffs
RoboFinSystems/robosystems
RoboSystems is a financial knowledge graph platform that transforms complex financial data into...