GitHub Repository

Agent skill that routes each coding task to the most token-efficient tool per layer: serena for code reads, rtk for command output, caveman for prose, Ponytail for generated code — −70% tokens measured in a 9-tool benchmark. Installable as a Claude Code plugin or Codex skill.

1 stars

Token-saviour – agent skill from benchmarking 9 token-saving tools

Name: Token-saviour – agent skill from benchmarking 9 token-saving tools
Availability: InStock
Author: vagkaratzas

by vagkaratzas·Jul 8, 2026·2 points·4 comments

Visit Project View on HN

AI Analysis

●●SolidBig BrainSolve My Problem

Benchmark-backed router cuts agent token costs by 70% across four task layers.

Strengths

•Benchmarks nine tools to assign winners for reading, writing, and command tasks.
•Installable as native Claude Code plugin or Codex skill with zero config.
•Layer-based routing logic prevents expensive tools from handling specific cheap tasks automatically.

Weaknesses

•Relies on external tools like serena and rtk staying maintained and compatible.
•Token savings vary wildly depending on task complexity and model pricing changes.