skills.rest
ax-llm
ax-llmOFFICIAL
Explore all skills by this author →
GitHub

ax-gepa

Generate AxGEPA optimization code for Ax programs, flows, and agents.

2.9k|186|Updated Feb 23, 2023
Search Tags:#typescript#optimization#agents#pareto#gepa#prompt tuning#axgepa
One-click install
npx skills add https://github.com/ax-llm/ax --skill ax-gepa-ax-llm
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: ax-gepa
Source: https://github.com/ax-llm/ax/tree/main/website/static/typescript/.well-known/agent-skills/ax-gepa
Command: npx skills add https://github.com/ax-llm/ax --skill ax-gepa-ax-llm

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Ax-GEPA helps you generate correct optimization code for Ax programs when you need to improve prompts, components, or agent trees without hand-tuning every instruction.

Core Features & Use Cases

  • GEPA optimization guidance: Write code for AxGEPA and the top-level optimize flow with the right defaults for student and teacher models.
  • Validation-aware tuning: Use train and validation examples, bounded metric calls, and deterministic scoring to select the best optimization candidate.
  • Advanced optimization workflows: Handle scalar or Pareto metrics, bootstrap few-shot refinement, tree-aware component maps, and saving or restoring optimized programs.
  • Use case: A developer wants to improve a classifier or multi-step flow and needs correct Ax code that applies optimization results back to the program safely and consistently.

Quick Start

Ask for AxGEPA optimization code for your Ax generator or flow, including a metric, validation examples, and the correct way to apply the optimized result.

Frequently Asked Questions about ax-gepa

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

FAQPage Schema
How do I optimize TypeScript agent prompts using GEPA?▼

AxGEPA generates optimization code for Ax programs by applying GEPA tuning, Pareto optimization, and bootstrap few-shot refinement to agent trees and flows, requiring deterministic metrics and separate training and validation sets to select the best optimization candidate.

What do I need to set up before running Pareto optimization on Ax flows?▼

Running Pareto optimization on Ax flows requires deterministic metrics for scoring, separate training and validation sets for evaluation, and bounded metric calls to control optimization costs during validation-driven selection.

Can I use bootstrap few-shot refinement for multi-step agent trees?▼

Yes, you can apply bootstrap few-shot refinement to multi-step agent trees using AxGEPA, which handles tree-aware component maps to optimize instructions across compiled target runtimes and serialize the optimized programs.

How do I apply optimized prompt results back to an Ax program safely?▼

You apply optimized prompt results safely by using AxGEPA's correct application or serialization logic, which restores validation-driven selection results to your Ax generator or flow while maintaining consistency across compiled target runtimes.

Why does my GEPA prompt tuning return inconsistent validation scores?▼

Inconsistent validation scores during GEPA prompt tuning occur when metric calls are not deterministic, violating the strict AxGEPA requirement needed to accurately perform validation-driven selection and select the best optimization candidate.

Related Skills

Explore More Software Engineering Skills→
ax-llmax-llm
2.9k

ax-java-agent-optimize

Configure Java Ax agent optimization runs with evaluators and bounded budgets.

Official
Intermediate
ax-llmax-llm
2.9k

ax-playbook

Generate Ax playbook code for evolving and applying context knowledge in TypeScript.

Official
Advanced
ax-llmax-llm
2.9k

ax-go-refine

Refine Go Ax generation workflows with evaluator feedback and optimizer APIs.

Official
Intermediate
drbobberdrbobber

prompt-engineering-expert

Analyze and refine AI prompts with advanced techniques like Chain-of-Thought.

Community
Advanced