Context-Engine

•
402
•

Enhances AI agents and integrated development environments with an open-core, self-improving context compression and code search suite.

Context-Engine is an open-core, agentic context compression suite designed to revolutionize how AI agents and integrated development environments (IDEs) interact with codebases. It moves beyond traditional large file chunks to provide highly precise 5-50 line code spans using hybrid search with semantic, lexical, and cross-encoder reranking. This self-hosted Docker stack ensures no cloud dependency and features adaptive learning, continually improving its search capabilities with every use. Context-Engine offers universal compatibility with various AI clients and IDEs like Claude Code, Cursor, and Windsurf, along with a robust memory system for storing and retrieving team knowledge.

Características Principales

01Precise micro-chunking for highly relevant code retrieval (ReFRAG-inspired).
02Integrated memory system for storing and retrieving team knowledge.
03Adaptive learning for continuous search improvement based on usage.
04Self-hosted Docker stack with no cloud dependency.
05276 GitHub stars
06Universal compatibility with numerous AI clients and IDEs via MCP.

Casos de Uso

01Enhancing AI agents and LLMs with precise, context-aware code and memory retrieval.
02Establishing a self-hosted, adaptive knowledge base for development teams.
03Providing intelligent, self-improving code search capabilities directly within IDEs like VS Code.