The open-source coding agent Maki, developed by tontinton/maki, aims to significantly reduce token consumption in AI-powered scripting. It achieves this through several innovative techniques, including an index tool that uses tree-sitter to parse files and only read necessary functions, a sandboxed interpreter for executing tools asynchronously, and the ability to delegate tasks to different model strengths like Anthropic's Haiku, Sonnet, or Opus. Maki also integrates with tools like rtk to shrink bash output, further minimizing token usage. AI
IMPACT Maki's innovative approach to token efficiency could significantly lower operational costs for AI-powered scripting and development.
RANK_REASON The item describes a new open-source tool designed to optimize existing AI workflows, rather than a core AI model release or research.
Read on dev.to — Claude Code tag →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →