Welcome! 👋 This repository showcases a specialized set of instructions, prompt.md, designed to empower AI agents (like those in Cursor or similar environments) to perform complex tasks with enhanced systematism, predictability, and effectiveness.

The primary goal is to demonstrate and refine a detailed "persona" or rule set that guides an AI agent acting as a versatile AI Minion. This is achieved by providing a comprehensive playbook, prompt.md, which defines a specific role, a rigorous workflow, and precise interaction patterns for tackling various assignments. 🧠

The cornerstone of this repository is prompt.md. This document meticulously outlines a strict, step-by-step process for the AI minion, covering:

• Requirement Analysis: Understanding stakeholder needs and performing initial clarity checks.
• Context & System Review: Thoroughly investigating existing systems, codebase (if applicable), or relevant information.
• Interactive Q&A: Formulating precise questions to resolve ambiguities with the stakeholder (user).
• Detailed Action Planning: Proposing a granular, verifiable plan before any execution.
• Sequential & Secure Execution: Implementing tasks methodically, with a strong emphasis on security (where relevant) and quality.
• Comprehensive Verification: Ensuring all work meets requirements and quality standards.

prompt.md is built upon a set of Core Operational Principles, emphasizing:

• Systematic Execution
• Proactive Clarification & Explicit-First Approach
• Comprehensive Security by Design (especially for software/data tasks)
• Contextual Awareness
• Extend by Default (for existing systems/code)
• Strict Plan Adherence
• Verifiability & Testability
• Professional and Precise Communication

This prompt is an example of advanced prompt engineering, designed to enable an AI agent to tackle complex assignments in a structured and reliable manner. For a detailed, step-by-step illustration of this prompt in action, please see example-conversation.md.

The process defined in prompt.md can be visualized as follows:

graph TD
    A[Stakeholder Request] --> B(Analyze Request <br> & Setup Plan);
    B --> C{Initial Clarity?};
    C -- Yes --> D[Analyze Context <br> /Codebase];
    C -- No/Critical Questions --> E[Ask Questions];
    D --> E;
    E --> F{All Questions <br> Answered?};
    F -- Yes --> G[Prepare Action Plan];
    F -- No --> E;
    G --> H{Plan Approved?};
    H -- Yes --> I[Execute Tasks <br> Sequentially];
    H -- No --> G;
    I --> J{All Tasks Done <br> & Verified?};
    J -- Yes --> K[Validate Plan Completion <br> & Report];
    J -- No --> I;
    K --> L[Output: Completed Work <br> & Documentation];

    subgraph Iterative Clarification & Planning
        D
        E
        F
        G
        H
    end

    subgraph Execution Loop
        I
        J
    end

The prompt.md is designed to be used with advanced AI assistants to guide them in performing complex tasks. Due to its comprehensive nature, specific setup approaches are recommended for optimal use:

• GitHub Copilot:

  • Option 1 (Simplest - Direct Prompting): Paste the full prompt.md content directly into the chat for specific complex tasks.
    • See: Copilot: Guide to Direct Prompting with prompt.md
  • Option 2 (General Guidance - Instructions File): Use a concise .github/copilot-instructions.md file for repository-level hints, complementing direct prompting.
    • See: Copilot: Using the .github/copilot-instructions.md File

• Cursor:

  • Option 1 (Simplest - Direct Prompting): Paste the full prompt.md content (or use @prompt.md) directly into the chat.
    • See: Cursor: Guide to Direct Prompting with prompt.md
  • Option 2 (General Guidance - Rules File): Use a concise .cursorrules file for project-level hints.
    • See: Cursor: Using the .cursorrules File for General Guidance
  • Option 3 (Integrated Persona - Custom Mode): Create a dedicated "Custom Mode" embedding the full prompt.md.
    • See: Cursor: Creating a "Minion" Custom Mode

These documentation files provide detailed steps for each approach.

Working with an AI agent powered by prompt.md is a highly interactive and collaborative dialogue 💬. This ensures the AI fully understands your needs and provides full transparency into its process.

The interaction model is as follows:

1. Your Request, Its Plan: You provide a requirement 📝. The AI Minion, following prompt.md, first creates a dedicated plan file (e.g., in .minions/plans/your_plan_title.md). This file becomes the central hub 🏡 for the entire task.
2. Deep Dive & Intelligent Questions: The agent meticulously analyzes your request 🕵️‍♂️, examines the existing context (codebase, documents, etc.), and identifies potential ambiguities or risks. It then populates the plan file with specific, contextual questions 🤔. This is an intelligent feedback mechanism to avoid misunderstandings.
3. Your Insights are Key: You, as the stakeholder, review these questions directly in the plan file and provide answers 🗣️. This iterative Q&A is crucial for shared understanding.
4. Transparent Action Planning: Once clarity is achieved, the agent proposes a detailed, step-by-step action plan 🗺️ within the same plan file.
5. Your Green Light: You review this action plan. The AI agent waits for your explicit approval 👍 before executing tasks.
6. Sequential Execution & Reporting: Upon approval, the agent works through tasks sequentially ⚙️, updating a corresponding .tasks.md file with progress.
7. Final Review & Handoff: Once all tasks are complete, the agent provides an "Implementation Summary" and "Testing Notes" (or equivalent documentation) 📄 in the plan file.

This structured interaction offers:

• Reduced Misunderstandings: Proactive questioning minimizes guesswork ✔️.
• Increased Transparency: You always know the AI's plan and progress 🔎.
• Greater Control: Key decisions remain with you 🕹️.
• Higher Quality & Secure Outcomes: Addressing issues early ensures the result meets precise requirements and relevant standards 🎯.

Think of this as partnering with a diligent, systematic AI minion who prioritizes clarity, and methodical execution 📏✂️.

A detailed, step-by-step example of this interaction can be found in example_conversation.md.