prompt-library-minimal_skill

Overview

This skill provides a library of minimal, token-efficient prompts and templates designed to produce reliable results while reducing API cost and latency. It focuses on concise, production-ready prompts for common developer tasks like code review, bug fixing, generation, refactoring, testing, and documentation. The goal is repeatable savings and faster responses without sacrificing output quality.

How this skill works

The library gives short imperative templates and examples that replace long natural-language requests with focused instructions and required fields. It supplies format and constraint patterns (JSON, code-only, length limits) and combines them with best-practice heuristics like specifying output shape and removing pleasantries. Copy a template, fill the placeholders, and send it to your model to get consistent, low-token responses.

When to use it

• When you need to reduce token costs and latency in production
• When you want predictable, repeatable prompt outputs
• When sending high-volume requests (CI, bots, batch jobs)
• When you require strict output formats (JSON, code-only)
• When you want compact templates for developer workflows

Best practices

• Write commands in imperative mood (e.g., "Refactor for readability")
• Specify exact output format and constraints (fields, length, language)
• Include only essential context and a minimal example or code snippet
• Omit pleasantries and filler words to save tokens
• Standardize templates across your team for repeatability

Example use cases

• Automated code review: use "Review for bugs, performance, security" plus code snippet
• Bug triage: "[Issue] fails with [error] Code: [snippet] Expected: [behavior]"
• Generate functions: "Function: sum two numbers Include: error handling, comments Language: TS"
• Refactor tasks: "Refactor for readability, efficiency: [code]"
• Testing scaffolds: "Jest tests (normal, edge, error): [code]"