oimiragieo/agent-studio

token-saver-adaptive-ratio

Helper skill that exposes adaptive token-saver compression ratios for large context corpora.

查看源码

使用方法

  1. 复制安装命令,在自己的智能体环境安装前先检查源码包。
  2. 从原始文档核对所需工具、凭据和支持的宿主。
  3. 说明输入材料与期望交付物,并对照来源检查输出。
npx skills add https://github.com/oimiragieo/agent-studio --skill token-saver-adaptive-ratio

开始前需了解

  • 安装不包含模型订阅、API 额度或访问文件的权限。
  • 仓库星数和安装量反映活跃情况,不等于独立质量或安全测试。
仓库原始内容

按源仓库内容呈现,保留标题、案例、代码、表格、链接以及原文引用的演示图片。

Token Saver Adaptive Ratio

Computes adaptive skeleton ratio based on corpus token count for context compression.

Purpose

This skill provides the computeAdaptiveRatio function that determines the optimal compression ratio based on the size of the context corpus:

  • <8K tokens: Returns 0.8 (keep 80% of content)
  • 8-32K tokens: Returns 0.5 (keep 50% of content)
  • 32-100K tokens: Returns 0.2 (keep 20% of content)
  • >100K tokens: Returns 0.1 (keep 10% of content)

Usage

javascript
const { computeAdaptiveRatio } = require('./scripts/main.cjs');

const ratio = computeAdaptiveRatio(50000); // Returns 0.2

When to Use

  • Before running context compression to determine the target compression ratio
  • When estimating token budget for large file reads
  • As part of the token-saver-context-compression workflow

Implementation

Re-exports the computeAdaptiveRatio function from token-saver-context-compression for module resolution and testing purposes.