Do I need a dataset to use the optimization workflow?

No, while datasets help with generalization, the skill supports single-task optimization against a single evaluator for targeted improvements.

How can I improve the quality of the optimization results?

The best way to improve results is by providing rich feedback in your evaluator output, including sub-scores and specific improvement hints that the engine can use for reflection.

What kind of artifacts can I optimize with this skill?

You can optimize any text-based artifact, including system prompts, code snippets, configuration files, and even other specialized Claude Code skills.

How does early stopping work in the optimization process?

The system automatically monitors score progress and triggers early stopping if performance plateaus, preventing unnecessary iterations and saving compute budget.

Artifact Optimization Guide

Name: Artifact Optimization Guide
Author: ASRagab

byASRagab

0•

Data Science & ML

Optimizes text artifacts including prompts, code, and configurations through automated evolutionary feedback loops.

The Artifact Optimization Guide provides a comprehensive framework for iteratively improving text-based assets using the gepa engine. It guides users through preparing seed artifacts, creating custom evaluators, and configuring optimization parameters like budget and early stopping to maximize quality across single-task or multi-task datasets. This skill is essential for developers looking to fine-tune LLM prompts, refactor complex configurations, or refine specialized code skills where performance metrics and objective-based scoring are critical for success.

Key Features

010 GitHub stars

02Support for single-task, multi-task, and generalization datasets

03Detailed result interpretation with score progression tracking

04Automated evolutionary optimization for text artifacts

05Integrated disk caching for performance across multiple runs

06Configurable optimization budgets and early-stopping mechanisms

Use Cases

01Optimizing software configuration files for performance or security constraints

02Iteratively improving specialized Claude Code skills based on evaluator feedback

03Refining LLM system prompts for better accuracy and instruction following

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add asragab/optimize-anything optimization-guide

For use in Claude.ai and ChatGPT

Key Features

010 GitHub stars

02Support for single-task, multi-task, and generalization datasets

03Detailed result interpretation with score progression tracking

04Automated evolutionary optimization for text artifacts

05Integrated disk caching for performance across multiple runs

06Configurable optimization budgets and early-stopping mechanisms

Use Cases

01Optimizing software configuration files for performance or security constraints

02Iteratively improving specialized Claude Code skills based on evaluator feedback

03Refining LLM system prompts for better accuracy and instruction following

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add asragab/optimize-anything optimization-guide

For use in Claude.ai and ChatGPT