skills.rest
kwakseongjae
kwakseongjaeCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

omd-lab-02-design-harness

Compare design harness versions by quality, token cost, iteration count, and abandonment rate.

416|35|Updated Apr 13, 2026
Search Tags:#performance evaluation#version comparison#design refinement#AI coding agents#design harness
One-click install
npx skills add https://github.com/kwakseongjae/oh-my-design --skill omd-lab-02-design-harness
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: omd-lab-02-design-harness
Source: https://github.com/kwakseongjae/oh-my-design/tree/main/.claude/skills/omd-lab-02-design-harness
Command: npx skills add https://github.com/kwakseongjae/oh-my-design --skill omd-lab-02-design-harness

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill helps refine design harnesses for AI coding agents by comparing different versions of the same task, analyzing quality, token cost, iteration count, and persona abandonment rate.

Core Features & Use Cases

  • Version Comparison: Run the same task with different harness settings (v1, v2, v3, etc.) and compare the outcomes.
  • Quality Analysis: Evaluate the quality of the outputs across different versions.
  • Token Cost & Iteration Count: Assess the token cost and iteration count for each version.
  • Persona Abandonment Rate: Analyze the persona abandonment rate for each version.
  • Use Case: For a product design task, you can use this Skill to compare the design outputs with different harness settings to determine the most effective configuration.

Quick Start

Run the design harness for the 'product design task' with version v2 and compare it with v1.

Frequently Asked Questions about omd-lab-02-design-harness

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

FAQPage Schema
How do I compare different versions of an AI coding agent design harness?▼

Compare design harness versions by running the same task with different configurations and evaluating quality, token cost, iteration count, and persona abandonment rate to determine the most effective setup.

How do I refine a design harness configuration for an AI coding agent?▼

Refine a design harness by executing iterative design tasks across multiple versions, analyzing output quality and persona abandonment metrics to isolate the optimal configuration settings.

What metrics should I track when evaluating AI coding agent design outputs?▼

Track output quality, token cost, iteration count, and persona abandonment rate to comprehensively evaluate and compare the performance of different AI coding agent design harness versions.

Can I use this design harness to compare configurations for a specific product design task?▼

Yes, you can run a specific product design task with different harness settings like v1 and v2 to directly compare design outputs and identify the most effective configuration.

What is the best way to manage version control for iterative AI design tasks?▼

Manage version control for iterative design tasks by systematically running and logging the same task with different harness settings, allowing direct performance comparison across versions.

Why does my AI coding agent abandon its persona during design iterations?▼

Analyze persona abandonment rates across different harness versions to identify configuration settings that cause the AI coding agent to drop its assigned persona during design iterations.

Related Skills

Explore More Design & Creative Skills→
yu13130122297yu13130122297

scout

Identify uncertainties in research framing, datasets, metrics, and baselines.

Community
Intermediate
m16khbm16khb
1

self-verify

Execute a 95-point QA gate and analyze JSON output from self-verification commands.

Community
Advanced
nbarthelemynbarthelemy

autonomous-loop

Manage autonomous iterative development loops with checkpointing, cost tracking, and safety limits.

Community
Advanced
iabheejitiabheejit

desloppify

Scan codebases for technical debt and quality issues.

Community
Advanced