Can I use this skill to optimize non-code assets?

Yes, it supports domains like 'marketing' and 'content' and includes evaluators for things like copy engagement and headline quality scores.

What is the primary purpose of the /ar:setup command?

It is used to initialize a new optimization experiment by collecting essential data like the target file, the evaluation command, and the success metrics.

How does the interactive mode help during setup?

The interactive mode prompts you step-by-step for the domain, target file, and metric, ensuring that the target file exists and the configuration is valid before starting.

What built-in evaluators are available?

The skill includes evaluators for benchmark speed, file size, test pass rates, build speed, memory usage, and several LLM-based quality judges.

Where does the skill store my experiment data?

During setup, you can choose to store data either within your project's .autoresearch/ folder or in your global user directory (~/.autoresearch/).

Autoresearch Experiment Setup

Name: Autoresearch Experiment Setup
Author: JantonioFC

byJantonioFC

•

分析与监控

Configures automated research experiments to optimize code performance, file size, and content quality through systematic benchmarking.

The Autoresearch Experiment Setup skill provides a robust framework for initializing optimization tasks within Claude Code. It allows users to define specific domains, target files, and evaluation commands to create a repeatable environment for automated improvements. Whether you are tuning API performance, reducing bundle sizes, or perfecting LLM prompts, this skill guides you through selecting the right metrics and directions for success. By integrating built-in evaluators for speed, memory, and qualitative scoring, it bridges the gap between manual coding and autonomous performance engineering.

主要功能

01Support for LLM-based qualitative judging for content and prompts

02Flexible storage options for project-specific or global configurations

032 GitHub stars

04Interactive configuration wizard for multi-parameter experiment setup

05Automated baseline metric verification and experiment branching

06Built-in evaluators for benchmarking speed, size, and memory usage

使用场景

01Automating the optimization of API latency by benchmarking execution times

02Iteratively improving system prompts using an LLM-as-a-judge scoring system

03Reducing memory footprints and bundle sizes for resource-constrained applications

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add jantoniofc/skillsbank setup

For use in Claude.ai and ChatGPT