About

Context Mode is an MCP (Multi-Modal Chat Protocol) server designed to tackle efficiency and memory issues in large language models (LLMs) when managing context. It significantly reduces raw data in the context window via sandboxed tools, achieving up to a 98% context reduction. It ensures session continuity by tracking session states using SQLite and FTS5 for relevant data retrieval. The tool promotes an 'infer in code' paradigm, enabling LLMs to execute scripts for analysis instead of processing vast amounts of raw data, thereby saving context and reducing output verbosity by 65-75%, enhancing LLM interaction efficiency.

Features

• High Context Data Compression (up to 98%)
• SQLite/FTS5-based Session State Persistence & Continuity
• 'Infer in Code' Driven Analysis & Script Execution
• Automated LLM Output Trimming & Compression (65-75% token saving)
• Deep Integration with Major AI Dev Platforms (Claude Code, Gemini CLI, VS Code Copilot)

Supported Platforms