🚀 Advanced Prompts Factory

Enterprise-grade meta-prompts for GitHub projects

Quick Start • Documentation • Prompts • Contributing • License

---

🌟 Overview

Advanced Prompts Factory is a curated collection of production-ready meta-prompts designed to generate complete, enterprise-grade GitHub projects in a single step. Born from the need to eliminate generic placeholders and stack-agnostic templates, this factory provides intelligent, context-aware prompts that adapt to your technology stack while maintaining strict quality standards.

The Problem: Most project generators produce generic skeletons filled with [TODO], [Your description here], and placeholder code that doesn't run. Developers waste hours customizing templates manually, fighting with mismatched stack configurations, and fixing linter violations.

The Solution: Meta-prompts that perform silent stack analysis before generation, producing real, executable code tailored to your specific technology (PowerShell, Node.js, Python, Nix, etc.) with zero placeholders. Every command works out-of-the-box, every CI/CD pipeline runs correctly, every .gitignore matches your stack.

Key Benefits:

• ⚡ Zero-placeholder generation: All files contain real, functional content
• 🎯 Stack-aware intelligence: Automatic adaptation to 10+ technology stacks
• 🌍 Bilingual by design: EN/US + FR documentation in every project
• 🏆 Enterprise standards: markdownlint compliance, Conventional Commits, security policies
• 📦 15+ files per project: Complete GitHub ecosystem (CI/CD, templates, docs, security)

Target Audience: Solo developers, open-source maintainers, and enterprise teams who value quality, consistency, and speed in their project scaffolding.

---

✨ Features

• 🎯 RepoArchitect Pro: The flagship meta-prompt for generating complete GitHub repositories with stack-specific configurations, bilingual READMEs, and real CI/CD pipelines. Supports PowerShell, Node.js, Python, Nix, Bash, and more.

• ⚡ Silent Stack Analysis: Pre-generation intelligence layer that detects your project type (CLI tool, library, API, web app) and technology stack, then tailors every file accordingly—no user interaction required.

• 🛡️ Zero-Placeholder Guarantee: Every generated file contains executable code. Commands run as-is, tests pass, CI/CD deploys. No [TODO] comments, no [Setup steps] sections—only production-ready content.

• 🚀 One-Step Generation: Paste a meta-prompt into any LLM, describe your project in one sentence, and receive 15+ fully-formed files instantly. No iterative back-and-forth, no manual template customization.

• 💡 Extensible Architecture: Add your own meta-prompts to the collection following our standardized template. Each prompt includes version tracking, usage statistics, and stack compatibility matrices.

---

🚀 Quick Start

Prerequisites

• Git 2.30+ (for cloning and version control)
• LLM Access (Claude 3.5+, GPT-4, or compatible API)
• Text Editor (VS Code, Vim, or any markdown-compatible editor)
• Optional: PowerShell 7.6+ (for automation scripts)

Installation

# Clone the repository
git clone https://github.com/valorisa/Advanced-Prompts-Factory.git
cd Advanced-Prompts-Factory

# Browse available prompts
ls -l prompts/

# Copy your desired meta-prompt
cat prompts/repoarchitect-pro.md

# Paste into your LLM and describe your project
# Example: "Generate a PowerShell CLI tool for parsing log files"

Tip

Pro tip: Start with repoarchitect-pro.md for your first project. It's the most mature and battle-tested prompt in the collection, supporting 10+ technology stacks out-of-the-box.

---

📖 Documentation

Architecture Overview

Advanced Prompts Factory follows a modular, template-driven architecture where each meta-prompt is a self-contained specification document:

graph LR
    A[User Input: Project Description] --> B{Meta-Prompt Engine}
    B --> C[Phase 0: Silent Stack Analysis]
    C --> D[Stack Detection: Language + Type]
    D --> E{Decision Matrix}
    E --> F[Template Selection]
    F --> G[File Generation: 15+ artifacts]
    G --> H[Quality Checks: markdownlint + executable]
    H --> I[Output: Complete Project]
    
    style C fill:#e1f5ff
    style E fill:#fff3cd
    style H fill:#d4edda

Loading

Component Breakdown:

1. Meta-Prompt Repository (prompts/): Versioned collection of prompt templates, each with explicit stack support matrices
2. Stack Detection Layer: Embedded intelligence within prompts that identifies technology and project type
3. Template Engine: Conditional logic that swaps generic placeholders for stack-specific real code
4. Quality Assurance: Built-in checklists ensuring zero-placeholder delivery and linter compliance

Configuration Guide

Each meta-prompt can be customized via inline variables before pasting into your LLM:

# Inside any prompt file, locate the configuration section:

## Configuration Variables (Modify before use)
- **DEFAULT_LICENSE**: MIT | Apache-2.0 | GPL-3.0 | Proprietary
- **BILINGUAL_MODE**: true | false (disable French translations if not needed)
- **MIN_FILE_COUNT**: 15 (increase for more comprehensive projects)
- **AUTHOR_USERNAME**: valorisa (replace with your GitHub username)

Alternatively, override at generation time by appending instructions:

"Use Apache-2.0 license instead of MIT"

Troubleshooting

Problem: Generated CI/CD workflow fails with "command not found"
Solution: The LLM may have hallucinated a command. Cross-check the workflow against the official GitHub Actions marketplace for your stack's action (e.g., actions/setup-node@v4 for Node.js).

Problem: README contains [TODO] placeholders despite using zero-placeholder prompts
Solution: Your LLM may be outdated or context-limited. Try:

1. Use a more recent model (Claude 3.5 Sonnet or GPT-4 Turbo)
2. Split the generation into two passes: structure first, then content fill
3. Explicitly add "Replace ALL placeholders with real content" to your user message

Problem: .gitignore doesn't match my stack (e.g., Python project has Node.js ignores)
Solution: The stack detection failed. Be explicit in your project description:

❌ "Generate a data processing tool"
✅ "Generate a Python CLI tool using Click for CSV data processing"

Problem: Generated project is too verbose / too minimal
Solution: Adjust the verbosity instruction in the meta-prompt:

• For minimal: Change "README > 2000 words" to "README 500-800 words"
• For verbose: Add "Include inline code comments explaining complex logic"

Problem: Bilingual sections are missing or incomplete
Solution: The LLM ran out of context or tokens. Try:

1. Generate in two phases: EN-only first, then translate FR sections separately
2. Use a model with larger context (200K+ tokens)
3. Remove less critical files (e.g., docs/api.md) to free up generation budget

---

🛠️ Development

Adding a New Meta-Prompt

1. Create the prompt file:

   # From repository root
   New-Item -Path "prompts/my-new-prompt.md" -ItemType File

2. Follow the standard template:

   # Prompt Name
   ## Version: 1.0.0
   ## Stacks Supported: [list]
   ## Project Types: [list]
   
   [Your prompt content following RepoArchitect structure]

3. Add metadata entry to docs/prompts-index.md:

   | my-new-prompt | Description | v1.0.0 | Stack1, Stack2 | Type1, Type2 |

4. Test the prompt with at least 3 different project descriptions covering different stacks

5. Submit a PR following CONTRIBUTING.md

Project Structure

Advanced-Prompts-Factory/
├── prompts/                     # Meta-prompt collection
│   ├── repoarchitect-pro.md    # Flagship prompt
│   └── [future-prompts].md
├── docs/
│   ├── architecture.md          # System design documentation
│   ├── prompts-index.md         # Catalog of all prompts
│   └── usage-examples.md        # Real-world usage demonstrations
├── scripts/                     # Automation utilities
│   └── Make.ps1                # Build, test, validate commands
├── .github/
│   ├── workflows/
│   │   ├── ci.yml              # Markdown linting + link checking
│   │   └── release.yml         # Semantic versioning automation
│   └── ISSUE_TEMPLATE/
└── README.md                    # You are here

Running Validations Locally

# Load the Make.ps1 functions
. ./scripts/Make.ps1

# Validate all markdown files
Invoke-Lint

# Check for broken links
Invoke-LinkCheck

# Run full quality suite
Invoke-Test

---

🧪 Testing

# Run markdown linting
Invoke-Lint

# Check for broken internal/external links
Invoke-LinkCheck

# Validate all prompts have required metadata
Invoke-ValidatePrompts

# Full test suite
Invoke-Test

Testing Strategy:

• Linting: markdownlint-cli2 with .markdownlint.json config for zero-violation enforcement
• Link Checking: Automated verification of all URLs and internal references
• Metadata Validation: Ensures every prompt file includes version, stack support, and usage stats
• Manual Smoke Tests: Each prompt tested with 3+ real project descriptions before release

---

🚢 Deployment

This repository is documentation-only and doesn't require deployment. However, if you're integrating these prompts into an automated service:

1. API Integration: Wrap prompts as API endpoints using your preferred framework (Express, FastAPI, Flask)
2. Versioning: Use GitHub releases to tag stable prompt versions
3. CDN Distribution: Host prompts/ directory on a CDN for low-latency access
4. Analytics: Track usage with custom headers or query params to measure prompt effectiveness

Example Integration (Node.js):

const fs = require('fs');
const express = require('express');
const app = express();

app.get('/prompt/:name', (req, res) => {
  const prompt = fs.readFileSync(`./prompts/${req.params.name}.md`, 'utf8');
  res.type('text/markdown').send(prompt);
});

app.listen(3000);

---

🤝 Contributing

We welcome contributions! Please see CONTRIBUTING.md for detailed guidelines.

Quick checklist:

• New prompts follow the standard template structure
• All markdown passes markdownlint validation
• Prompts tested with 3+ different project types
• Metadata added to docs/prompts-index.md
• Commit messages follow Conventional Commits format

---

📝 Changelog

See CHANGELOG.md

---

🛡️ Security

See SECURITY.md

---

📄 License

MIT License

This project is open-source and free to use. Attribution appreciated but not required.

---

👥 Team

• valorisa — Author & Maintainer — GitHub

---

🎯 Available Prompts

Prompt Name	Description	Version	Stacks	Types
RepoArchitect Pro	Complete GitHub project generation with stack-aware intelligence	2.0.0	PowerShell, Node.js, Python, Nix, Bash, Shell	CLI, Library, API, Web App, Scripts

See docs/prompts-index.md for detailed comparison and usage examples.

---

❓ FAQ

Q: Can I use these prompts with any LLM, or are they specific to Claude/GPT-4?
A: The prompts are LLM-agnostic and tested with Claude 3.5, GPT-4, and compatible models. However, best results require models with 100K+ context windows and strong instruction-following capabilities. Smaller models (7B-13B parameters) may produce incomplete or placeholder-heavy output.

Q: Do I need to credit Advanced Prompts Factory when using these prompts?
A: No. The MIT license allows unrestricted use without attribution. However, if you found a prompt valuable, a star ⭐ on the repo helps others discover it!

Q: Why are some generated files still generic despite using RepoArchitect Pro?
A: Two common causes: (1) Your project description lacks stack details—be explicit about your language and tools. (2) Your LLM ran out of tokens mid-generation—try a model with larger context or split generation into phases.

Q: Can I modify a prompt to remove the bilingual requirement?
A: Absolutely. Locate the # Instructions de Génération section in any prompt and change 6. **Bilingue EN/FR** to 6. **English-only**. This reduces token usage by ~30-40%.

Q: How do I contribute a new prompt to the factory?
A: See CONTRIBUTING.md for the full process. In short: fork the repo, add your prompt to prompts/, update docs/prompts-index.md, test with 3+ examples, and submit a PR. We review within 48 hours.

Q: Is there a CLI tool to automate using these prompts?
A: Not yet, but it's on the roadmap! Track progress in issue #1. In the meantime, you can create a simple wrapper script using PowerShell or Bash to automate clipboard copying.

---

Made with ❤️ by valorisa