Context-Engine: Production-ready MCP retrieval stack for AI coding assistants. Hybrid code search (dense + lexical + reranker), ReFRAG micro-chunking, local LLM prompt enhancement, and dual SSE/RMCP endpoints. One command deploys Qdrant-powered indexing for Cursor, Windsurf, Roo, Cline, Codex, and any MCP client.

This repository provides the companion Jupyter Notebooks for the book Context-Engineering for Multi-Agent Systems. It technically details the progressive construction of a Context Engine moving from simple prompts to foundational concepts to a production-ready system semantic blueprint. The code is a guide to building Multi-Agent Systems (MAS)

MCP server that provides code context and analysis for AI assistants. Extracts directory structure and code symbols using WebAssembly Tree-sitter parsers with Zero Native Dependencies.

Database Freedom Platform - Mathematical search optimization for whatever database you already have. 27,000x faster than vector databases with SMT-powered search across 8+ database types. One-time 9-2999 vs 00-500/month recurring.

A Local Context Engine

A production-ready TypeScript MCP server that provides comprehensive project analysis, intelligent code search, dependency tracking, and coordinated multi-file editing capabilities.

Deterministic, offline-first agent sandbox with concept-graph memory. Reflex→retrieval→planning→filter for repeatable outputs. Stable baseline for v4 external reasoning.

SkyOne is the cognitive nucleus of the Sky Ecosystem — an AI-driven bootloader designed to coordinate modular memory, journaling, and orchestration across all Sky tiers. It handles context persistence, data flow management, and reflection pipelines to unify intelligent behavior across SkyServer and SkyWeb.

Perplexity-powered agent for finding information regarding any real estate property. Built during the Perplexity Hackathon.