Token Optimizer MCP

Example Use Cases

• 1

  Reduce token consumption when sending large or repeated text by compressing and caching payloads before including them in the model context.

• 2

  Accurate token accounting and analysis to decide whether to compress content and estimate tokens saved and compression ratios.

• 3

  Persistent, disk-backed caching (SQLite) with cache statistics and automatic cleanup to improve performance and reduce repeated token costs.

Description

Token Optimizer MCP is a Model Context Protocol server that reduces token usage through Brotli compression and a persistent SQLite cache. It uses tiktoken (GPT-4 tokenizer) for accurate token counts and provides tools to compress, cache, analyze, and retrieve text to save tokens. The server is implemented in TypeScript, runs offline with zero external dependencies, and includes cache statistics and automatic cleanup for production use. Typical benefits are significant token and size reduction, persistent reuse of compressed content, and easy integration with Claude Code and Claude Desktop.