Context-Engine-AI/Context-Engine

Context-Engine MCP - Agentic Context Compression Suite

What it solves

It provides AI coding assistants with deep semantic understanding of a codebase, solving the problem of limited context and poor navigation in large or multi-repo projects. It enables assistants to perform complex code searches, track symbol relationships, and maintain persistent memory across sessions.

How it works

Context Engine integrates with AI assistants (like Claude Code, Cursor, Codex, and Windsurf) via "skills"—custom instructions and MCP (Model Context Protocol) tools. It uses an MCP bridge to connect a local codebase to the Context Engine backend, which indexes the code to enable semantic search, symbol graph navigation, and pattern recognition. Users can connect their codebase via a VS Code extension or a CLI tool that can run as a background daemon to watch for changes.

Who it’s for

Developers who use AI coding assistants and want their AI to have a more precise, structural, and persistent understanding of their entire project and cross-repo dependencies.

Highlights

  • Symbol Intelligence: Navigate callers, callees, definitions, and subclasses via a symbol graph.
  • Coded Memory: Store and recall knowledge using memory_store and memory_find for persistence across sessions.
  • Advanced Search: Supports semantic search, pattern search for structural code patterns, and git history search.
  • Efficiency: Batch queries (batch_search, etc.) can reduce token usage by over 75%.
  • Cross-Repo Tracing: Ability to trace flows across multiple repositories.
  • Broad Integration: Compatible with a variety of assistants including Claude, Cursor, Codex, Windsurf, Augment Code, and Gemini.