harness-tuning

Tighten and deduplicate prompt and CODEX content in Markdown files.

Updated May 4, 2026

Search Tags:#automation #editing #deduplication #skill-management #tuning #prompt #codex.md

Ready to run

npx skills add https://github.com/xaionaro/.config-codex --skill harness-tuning

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: harness-tuning
Source: https://github.com/xaionaro/.config-codex/tree/main/skills/harness-tuning
Command: npx skills add https://github.com/xaionaro/.config-codex --skill harness-tuning

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It helps teams write, review, shorten, tighten, deduplicate, or compress any skill file, system prompt, subagent prompt, or CODEX.md — including when asked to "fix" or "improve" a prompt.

Core Features & Use Cases

Reduce prompt length without changing intent.
Deduplicate overlapping rules and guidelines to ensure consistency.
Enforce a consistent style across SKILL.md, system prompts, and CODEX.md, enabling faster, more reliable prompt authoring.
Real-world example: compress a long policy into clear, imperative rules without losing meaning.

Quick Start

Paste your prompt file and ask Harness Tuning to produce a tighter, deduplicated version.

Frequently Asked Questions about harness-tuning

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

FAQPage Schema

How do I reduce prompt length without changing the original intent?▼

To reduce prompt length without changing intent, you can tighten and deduplicate prompt content to maximize clarity and minimize token waste. This process removes redundant wording and overlapping rules while preserving the original meaning.

What is the best way to deduplicate overlapping rules in a SKILL.md file?▼

The best way to deduplicate overlapping rules in a SKILL.md file is to apply a tightening process that enforces consistent style and removes redundant guidelines. This ensures consistency across your skill files and system prompts without losing meaning.

Can I use this approach to compress a long policy into clear imperative rules?▼

Yes, you can compress a long policy into clear, imperative rules without losing meaning. By tightening the prompt content and deduplicating overlapping instructions, you transform verbose policies into concise, executable guidance for your system or subagent prompts.

Does tightening a system prompt require a specific file format?▼

Tightening a system prompt requires a YAML frontmatter with name and description fields, using the Markdown body as executable guidance. Optional resources can be stored in scripts, references, and assets directories to support the automation of the prompt tightening process.

When should I not use prompt compression on a CODEX.md file?▼

You should not use prompt compression on a CODEX.md file when the original verbose context contains critical nuances or edge-case examples that cannot be converted into clear, imperative rules without losing essential meaning or altering the intended behavior.