MCP server that gets Claude to 97.9% (188/192) on a real coding benchmark at -80% active tokens and -83% wall time, vs 78.3% plain. Structural code navigation + persistent memory engine. Works with every MCP client.
Token Savior is an MCP server for AI coding agents that combines structural code navigation, persistent project memory, and Bash command rewriting. It is intended to help coding agents work through repositories with less token use and shorter execution times, and its catalogue add-ons configure or guide its use.
This repository also configures its own agents. See what token-savior tells them →
MCP server Claude CodeCodexCursor +2
One of 2 in server.json
Optimized MCP server for AI coding agents. Structural codebase navigation + persistent memory engine — 97.9% accuracy at −80% tokens on tsbench. Runs locally from the token-savior-recall Python package. Needs 3 environment variables to run.
MCP server Claude CodeCodexCursor +2
One of 2 in server.json
Optimized MCP server for AI coding agents. Structural codebase navigation + persistent memory engine — 97.9% accuracy at −80% tokens on tsbench. Runs locally from the token-savior-recall Python package. Needs 3 environment variables to run.