This repository contains examples of meta prompts that can be used with different AI coding tools (such as Claude Code, GitHub Copilot, Windsurf, Cursor, etc.) to enhance software development workflows. Prompts included there are to be used to generate prompts for various aspect of software development life cycle such as planning, coding, reviewing, and so on.
All meta prompts included in this repository has been tested with Claude Code (claude.ai/code) using Haiku 4.5 and Sonnet 4.5 models. Both models perform well although commands generated by Sonnet are more detailed and better suite for execution in Claude Code. It is therefore recommended to test several available models and choose the one that works best for your use case. Please note, that usually model selected for command generation does not have to be the same as the one used for executing the command (commands generated with Sonnet works well then executed by Haiku).
The plan-execute/ folder contains meta-prompts for the Plan & Execute method of software development with AI Agents. This methodology provides a structured workflow that separates development into two distinct phases:
- Planning Phase - Comprehensive analysis and preparation before implementation
- Execution Phase - Implementation based on the generated plan
-
Three-Command Workflow:
plan- Generates comprehensive development plans with context analysis, open questions, and detailed execution stepsexecute- Implements selected steps or phases from a plan with interactive checkpointsplan-update- Updates existing plans when specifications change or clarifications are needed
-
Deliverable Increments - Breaking work into discrete, testable chunks that can be safely merged independently
-
Atomic Commits - Following conventional commit format for clear development history
-
Context-Aware - Plans include root cause analysis, relevant code references, and stakeholder questions
-
Tool-Agnostic - Can be adapted for Claude Code, GitHub Copilot, Cursor, Windsurf, and other AI coding tools
For detailed information about the Plan & Execute methodology, see Plan & Execute README.