skills.rest
Loringtonian
LoringtonianCOMMUNITY
Explore all skills by this author →
𝕏@loringtonGitHub

golden-evolver

Evolve golden few-shot examples through scoring loops and versioned JSON storage.

12|2|Updated Jun 8, 2026
Search Tags:#summarization#prompt engineering#delegation#classification#few-shot#example curation#versioned datasets
One-click install
npx skills add https://github.com/Loringtonian/second-brain-template --skill golden-evolver
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: golden-evolver
Source: https://github.com/Loringtonian/second-brain-template/tree/main/.claude/skills/golden-evolver
Command: npx skills add https://github.com/Loringtonian/second-brain-template --skill golden-evolver

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps you systematically improve the few-shot examples that power your agent workflows, so classification, summarization, extraction, and delegation become more accurate over time.

Core Features & Use Cases

  • Example evolution: Runs iterative curation cycles that compare model outputs, gather owner scores, and promote the best examples into a versioned golden set.
  • Task coverage: Supports multiple brain workflows including classification, summarization, keyword extraction, template creation, segmentation, and delegation prompts.
  • Version control for prompts: Tracks active sets, historical scores, and candidate mutations so you can keep improving without losing prior results.
  • Use case: A knowledge-base owner wants better routing examples for content classification, so the Skill tests candidate examples on real inputs and saves the strongest set for future runs.

Quick Start

Use golden-evolver to improve the examples for a target task by reviewing real outputs, scoring them, and saving the approved set as the new active version.

Frequently Asked Questions about golden-evolver

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I improve few-shot examples for prompt classification and summarization?▼

You can improve few-shot examples by running iterative curation cycles that compare model outputs, gather owner scores, and promote the strongest examples into a versioned golden set. This refines classification, summarization, and extraction accuracy over time.

What is the best way to maintain versioned datasets for AI workflow prompts?▼

Maintaining versioned datasets involves tracking active example sets, historical scores, and candidate mutations in JSON storage. This approach loads active examples from context files at prompt-build time, ensuring you keep improving without losing prior results.

Can I use prompt engineering delegation for knowledge-base routing scenarios?▼

Yes, prompt engineering delegation applies to knowledge-base routing scenarios. The system tests candidate examples on real inputs, scores the outputs, and saves the approved set as the new active version to enhance routing accuracy.

Does example curation support keyword extraction and template creation workflows?▼

Example curation supports keyword extraction and template creation workflows. It applies scoring loops to evaluate candidate mutations, allowing you to systematically evolve and save the best performing examples for future prompt-driven tasks.

Why do I need scoring loops to refine few-shot prompts?▼

Scoring loops are required to evaluate and compare model outputs against real inputs systematically. By gathering owner scores through these loops, the system identifies and promotes the best examples into a versioned golden set, directly improving prompt accuracy.

Related Skills

Explore More Software Engineering Skills→
xiaolaixiaolai
111

writing-prompts

Structure prompts with a five-layer Role, Context, Task, Constraints, and Output framework.

Community
Advanced
pachoroapachoroa

security-audit

Detect malicious patterns and generate Markdown/JSON security reports for installed skills.

Community
Advanced
Wike-CHIWike-CHI
2

crm

Create and manage contact databases in JSON or CSV files.

Community
Intermediate
Pixel-Process-UGPixel-Process-UG
1

senior-prompt-engineer

Engineer prompts for large language models with structured outputs and evaluation.

Official
Advanced