prompt-engineering

Guide teams to write agent instructions with frontmatter and references.

3|Updated Jul 17, 2024
One-click install
npx skills add https://github.com/Castrozan/.dotfiles --skill prompt-engineering-castrozan
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: prompt-engineering
Source: https://github.com/Castrozan/.dotfiles/tree/main/agents/skills/prompt-engineering
Command: npx skills add https://github.com/Castrozan/.dotfiles --skill prompt-engineering-castrozan

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps teams maintain accurate agent/skill instructions as code evolves, preventing documentation rot and misalignment between guidance and implementation.

Core Features & Use Cases

  • Frontmatter-driven metadata: Ensures each skill exposes a name and description that remain synchronized with the codebase.
  • Dynamic WHAT guidance: Separation of WHAT and HOW enables updates without rewriting core logic.
  • Reference-backed instructions: Encourages linking to current modules, references, and state to stay aligned with code changes.

Quick Start

Ask the AI to draft an up-to-date agent instruction set for the current repository state, including frontmatter with name and description and a body that explains how to apply changes.

Frequently Asked Questions about prompt-engineering

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

FAQPage Schema
How do I keep agent instructions current as code evolves?

To keep agent instructions current as code evolves, use a frontmatter schema with name and description, separate WHAT guidance from HOW logic, and link instructions to current codebase references to prevent stale guidance.

What causes documentation rot in agent instructions and how to fix it?

Documentation rot occurs when agent instructions lose alignment with evolving code requirements. Fix it by applying reference-backed instructions that link to current modules, templates, and decision patterns via dynamic discovery.

How do I write agent instructions that separate WHAT and HOW?

Write agent instructions by defining the WHAT as dynamic guidance that updates without rewriting core logic, and the HOW as implementation details. Enforce a frontmatter schema exposing a name and description for each skill.

Does my codebase need a frontmatter schema for agent instructions?

Your evolving codebase needs a frontmatter schema if agent behavior must reflect current requirements. The schema ensures each skill exposes a synchronized name and description to prevent documentation misalignment.

When should I use reference-backed instructions instead of static agent guidance?

Use reference-backed instructions instead of static guidance when your codebase evolves frequently. Linking to current modules, templates, and state ensures agent behavior reflects current requirements and prevents stale decision patterns.