refine

Apply minimal updates to Codex skill metadata, workflows, and resources.

66|2|Updated Feb 18, 2015
One-click install
npx skills add https://github.com/tkersey/dotfiles --skill refine-tkersey
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: refine
Source: https://github.com/tkersey/dotfiles/tree/main/codex/skills/refine
Command: npx skills add https://github.com/tkersey/dotfiles --skill refine-tkersey

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Refine a Codex skill by applying focused, minimal updates to its metadata, workflow, and resources, turning scattered evidence into coherent improvements that enhance reliability and performance.

Core Features & Use Cases

  • Evidence-driven updates: transforms usage feedback, errors, and session-mining insights into concrete changes to SKILL.md, triggers, and resources.
  • Minimal-diff changes: prioritizes small, safe edits that preserve existing behavior and enable easy review.
  • Validation-oriented workflow: regenerates or updates agents/openai.yaml as needed and runs quick_validate to ensure quality before deployment.

Quick Start

Run refine on a target skill to apply the suggested updates and then validate with quick_validate.

Frequently Asked Questions about refine

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

FAQPage Schema
How do I apply minimal updates to Codex skill metadata without breaking existing workflows?

To apply minimal updates to Codex skill metadata safely, use a refine workflow that prioritizes small, focused diffs to SKILL.md and triggers. This preserves existing behavior while turning scattered feedback into coherent improvements.

What is the best way to validate Codex skills after modifying frontmatter and resources?

The best way to validate Codex skills after modifying frontmatter is to run an automated validation check using quick_validate. This ensures strict frontmatter conformance and verifies quality before deployment.

How does evidence-driven skill refinement work with session mining observations?

Evidence-driven skill refinement works by transforming usage feedback, errors, and session-mining insights into concrete changes. It analyzes these observations to apply targeted updates across triggers, descriptions, and assets.

When do I need to regenerate agents/openai.yaml during a skill update?

You need to regenerate agents/openai.yaml during a skill update when changes to metadata or resources require strict frontmatter conformance. The refinement workflow handles this regeneration automatically as needed.

Can I update skill workflows in-place while keeping the original assets intact?

Yes, you can update skill workflows in-place with minimal-diff changes. This approach prioritizes small, safe edits that preserve existing behavior and enable easy review without overwriting original assets.

Why does my Codex skill validation fail after manual metadata edits?

Codex skill validation fails after manual metadata edits when strict frontmatter conformance is broken. Applying focused updates through an automated refinement process and running quick_validate resolves these gaps.