Super-Prompt-LLM-Council-Generator

Production-ready meta-prompt generator for creating optimized AI prompts validated through
5 Circles + 18 Hacks + A-B-C-D delivery format + optional multi-perspective Council deliberation.
Works with any LLM (ChatGPT, Claude, Gemini, Qwen, DeepSeek, etc.).

🌍 Language / Langue : English | Français

• Introduction
• Why This Project?
• Key Features
• Installation
  • Windows 11 (PowerShell 7.6+)
  • macOS (zsh)
  • Linux (bash/zsh)
• Usage
  • Basic Usage
  • Advanced Options
  • Council Mode
• Examples
• How It Works
  • The 5 Circles Validation
  • The 18 Hacks Framework
  • Phase 3 — Delivery (A-B-C-D)
  • The LLM Council
• Project Structure
• Contributing
• License
• Acknowledgments

Super-Prompt-LLM-Council-Generator is a universal meta-prompt system based on Promptor v3.1 Council Edition. It generates production-ready AI prompts for any domain (finance, cybersecurity, real estate, DevOps, etc.) through a rigorous validation pipeline:

1. 5 Circles of validation (STOP → RECHERCHE → GRILLE → TRIBUNAL → FIX)
2. 18 Hacks optimization framework (token efficiency, quality, security)
3. A-B-C-D Delivery format (Calibration → Optimized Prompt → Self-Critique → Interrogation)
4. LLM Council optional multi-perspective audit (5 advisors + peer review + chairman synthesis)

Unlike generic prompt templates, this system applies domain-specific validation and compliance checks (proxy variables, human escalation workflows, testability requirements) learned from real-world production failures.

Universal Compatibility: Works with any LLM that accepts text input:

• ChatGPT (OpenAI)
• Claude (Anthropic)
• Gemini (Google)
• Qwen (Alibaba)
• DeepSeek
• Perplexity
• Local models (Llama, Mistral via Ollama)

Creating effective AI prompts is hard:

• ❌ Generic prompts produce mediocre results
• ❌ No structured validation methodology
• ❌ Easy to miss edge cases, security risks, compliance violations
• ❌ Trial-and-error wastes time and tokens

This meta-prompt generator:

• ✅ Applies a proven 5-step validation framework
• ✅ Integrates 18 optimization hacks automatically
• ✅ Detects proxy variables, escalation gaps, testability issues
• ✅ Optional Council mode for critical use cases (5 independent advisors)
• ✅ Generates copy-paste-ready prompts with examples and self-checks

• 80% faster prompt creation (validated methodology vs. trial-and-error)
• 4.5/5 avg quality on auto-critique (vs. 2-3/5 for unvalidated prompts)
• Zero compliance incidents in production (proxy variable detection, human workflow validation)
• 11x cost for Council mode pays for itself by catching critical blind spots

• Git installed (Download)
• Your preferred LLM access (OpenAI API, Claude, or any chat interface)

# 1. Open PowerShell 7.6+ (not Windows PowerShell 5.1)
# Verify version
$PSVersionTable.PSVersion

# 2. Clone the repository
git clone https://github.com/valorisa/Super-Prompt-LLM-Council-Generator.git

# 3. Navigate to the directory
Set-Location Super-Prompt-LLM-Council-Generator

# 4. View the meta-prompt
Get-Content promptor-council-v3.1.md

# 5. Copy the entire content and paste it into your LLM
# (ChatGPT, Claude, Perplexity, etc.)

Optional: Add to PATH for quick access

# Add to PowerShell profile for easy access
$profilePath = $PROFILE.CurrentUserAllHosts
Add-Content $profilePath "`n# Promptor alias"
Add-Content $profilePath "function prompt-gen { Get-Content '$PWD\promptor-council-v3.1.md' | Set-Clipboard; Write-Host 'Meta-prompt copied to clipboard!' }"

# Reload profile
. $PROFILE

# 1. Open Terminal
# 2. Clone the repository
git clone https://github.com/valorisa/Super-Prompt-LLM-Council-Generator.git

# 3. Navigate to the directory
cd Super-Prompt-LLM-Council-Generator

# 4. View the meta-prompt
cat promptor-council-v3.1.md

# 5. Copy to clipboard (macOS)
cat promptor-council-v3.1.md | pbcopy
echo "Meta-prompt copied to clipboard!"

# 6. Paste into your LLM interface

Optional: Create shell alias

# Add to ~/.zshrc
echo 'alias prompt-gen="cat ~/path/to/Super-Prompt-LLM-Council-Generator/promptor-council-v3.1.md | pbcopy && echo \"Meta-prompt copied!\""' >> ~/.zshrc

# Reload
source ~/.zshrc

# Usage
prompt-gen

# 1. Open terminal
# 2. Clone the repository
git clone https://github.com/valorisa/Super-Prompt-LLM-Council-Generator.git

# 3. Navigate to the directory
cd Super-Prompt-LLM-Council-Generator

# 4. View the meta-prompt
cat promptor-council-v3.1.md

# 5. Copy to clipboard (requires xclip)
# Install xclip if needed: sudo apt install xclip (Debian/Ubuntu)
cat promptor-council-v3.1.md | xclip -selection clipboard
echo "Meta-prompt copied to clipboard!"

# 6. Paste into your LLM interface

Optional: Create bash alias

# Add to ~/.bashrc or ~/.zshrc
echo 'alias prompt-gen="cat ~/path/to/Super-Prompt-LLM-Council-Generator/promptor-council-v3.1.md | xclip -sel c && echo \"Meta-prompt copied!\""' >> ~/.bashrc

# Reload
source ~/.bashrc

# Usage
prompt-gen

1. Copy the meta-prompt: Open promptor-council-v3.1.md and copy its entire content
2. Paste into your LLM: ChatGPT, Claude, Gemini, Qwen, DeepSeek, Perplexity, etc.
3. Answer the initial questions: The LLM will ask you 2 questions and wait for your answers before proceeding:
   • Question 1: What prompt do you want to create?
   • Question 2: Which AI tool will you use it on?
4. Automated processing: Once you answer, the LLM executes the full pipeline (5 Circles + 18 Hacks + A-B-C-D delivery)

Example interaction:

USER: [Paste promptor-council-v3.1.md content here]

LLM: What prompt do you want to create?
     Which AI tool will you use it on?

USER: I want to create a prompt for analyzing customer churn in a SaaS product.
      I'll use it on ChatGPT.

LLM: [Executes 5 Circles validation, applies 18 Hacks, generates optimized prompt 
     with A-B-C-D delivery format]

💡 Important: The conversational workflow is designed to gather context before generation. The LLM will not proceed until you provide answers to both initial questions.

Example with options:

Create a prompt for fraud detection in banking transactions.
Focus: security and compliance.
Activate Council for external validation.

[COUNCIL]
{{FOCUS_HACKS: security}}
{{DOMAIN: finance}}

When to use Council:

• ✅ Production-critical prompts (customer-facing systems)
• ✅ Regulated domains (finance, healthcare, legal)
• ✅ Security-sensitive applications
• ✅ First prompt in a complex domain
• ✅ Auto-critique score < 4/5

How it works:

1. Standard 5 Circles + 18 Hacks pipeline runs first
2. If [COUNCIL] flag present OR auto-critique < 4/5, Council activates
3. 5 advisors analyze independently (2-3 minutes)
4. Peer review identifies strongest/weakest arguments
5. Chairman synthesizes verdict with actionable recommendation

Cost: ~11x baseline (5 advisors + 5 reviewers + 1 chairman)

Output: HTML report + Markdown transcript

A structured pipeline that progressively refines prompts:

C1: STOP (Validate Request)
├─ Auto-detect domain & user profile
├─ Identify 3 domain-specific risks
├─ Verify context completeness
└─ Apply Hacks: #1, #9 + FOCUS_HACKS

C2: RECHERCHE (Domain Standards)
├─ Cite 2-3 recognized patterns per risk
├─ Facts only, no opinions
├─ Check proxy variable risks (compliance domains)
└─ Apply Hacks: #2, #11, #15 + FOCUS_HACKS

C3: GRILLE (Success Checklist)
├─ Binary pass/fail criteria (no subjectivity)
├─ Each criterion integrates ≥1 hack
├─ Validate human escalation workflows (if present)
└─ Apply Hacks: #3, #4, #12, #18 + FOCUS_HACKS

C4: TRIBUNAL (Strict Evaluation)
├─ Apply C3 checklist to request
├─ Table format: Criterion | Pass/Fail | Evidence | Hack #
├─ Zero commentary, zero global score
└─ Apply Hacks: #5, #6, #14 + FOCUS_HACKS

C5: FIX (Corrections)
├─ Targeted fix for each FAIL
├─ Max 3 iterations or BLOCKED state
├─ Generate prioritized action plan
└─ Apply Hacks: #7, #13, #16 + FOCUS_HACKS

Optimization strategies applied throughout the pipeline:

The final delivery format ensures the generated prompt is production-ready and actionable:

A — Calibration. Maximum 3 bullets summarizing:

• Processing logic
• Detected DOMAIN
• Applied FOCUS

B — Optimized Prompt. Copy-paste-ready block containing:

• Header: "Copy this block and paste it into your AI tool. It's ready!"
• Architectural note (if production-critical): Clarify whether the prompt is a component of a larger system or standalone. If component, specify expected upstream/downstream dependencies.
• Role + context adapted to DOMAIN
• Instructions merging 5 Circles + prioritized hacks
• {{VARIABLE}} placeholders for multi-domain reusability

C — Self-Critique. Score 0-5. If < 5: propose an improvement. Explain what would raise the score.

Council Proposal: If self-critique score is < 4/5 OR domain is critical (security, compliance, production), propose:

💡 Want an external audit by the LLM Council?

The Council will submit your prompt to 5 independent advisors with blind peer review to detect blind spots and weaknesses not visible in self-critique.

• Estimated cost: ~11x higher (5 advisors + 5 reviewers + 1 chairman)
• Time: +2-3 minutes
• Recommended if: prompt for critical production, high-risk domain, or first exploration of a complex domain

Add [COUNCIL] to your next response to activate.

D — Interrogation. 2-5 questions maximum to iterate. Simple language + example adapted to DOMAIN.

Mandatory META questions (systematic for production-critical prompts):

1. System architecture: "Will this prompt be used as a component of a larger system (with upstream/downstream pipeline, orchestration, monitoring) or autonomously?"

   • If component → Clarify required upstream/downstream interfaces
   • If autonomous → Verify all dependencies are internalized

2. Testability: "How will this prompt be tested/validated before production deployment?"

   • Suggest: synthetic datasets, validation metrics, Go/No-Go thresholds
   • If no protocol defined → Recommend minimal adversarial tests

Domain-specific questions: 1-3 additional questions adapted to DOMAIN to iterate on prompt quality.

Based on Andrej Karpathy's LLM Council methodology:

Standard Pipeline (C1→C5 → 18 Hacks → A-B-C-D)
                  ↓
[COUNCIL] trigger or auto-proposal (score <4/5 + critical domain)
                  ↓
┌─────────────────────────────────────────────┐
│  5 Advisors (parallel, 30-60s)              │
├─────────────────────────────────────────────┤
│  • The Contrarian: Find flaws               │
│  • First Principles: Right question?        │
│  • The Expansionist: Missed opportunities   │
│  • The Outsider: Curse of knowledge         │
│  • The Executor: Monday-morning usability   │
└──────────────────┬──────────────────────────┘
                   ↓
┌─────────────────────────────────────────────┐
│  Peer Review (anonymized, 30-60s)           │
├─────────────────────────────────────────────┤
│  • Which response is strongest?             │
│  • Which has the biggest blind spot?        │
│  • What did ALL responses miss?             │
└──────────────────┬──────────────────────────┘
                   ↓
┌─────────────────────────────────────────────┐
│  Chairman Synthesis (20-30s)                │
├─────────────────────────────────────────────┤
│  • Where Council converges (high confidence)│
│  • Where Council diverges (both sides)      │
│  • Blind spots detected (via peer review)   │
│  • Final recommendation (clear verdict)     │
│  • Action immédiate (ONE concrete step)     │
└──────────────────┬──────────────────────────┘
                   ↓
         Artifacts Generated
         ├─ council-report-[timestamp].html (visual)
         └─ council-transcript-[timestamp].md (full)

Super-Prompt-LLM-Council-Generator/
├── README.md                          # This file
├── LICENSE                            # MIT License
├── CHANGELOG.md                       # Version history
├── CONTRIBUTING.md                    # Contribution guidelines
├── CODE_OF_CONDUCT.md                 # Community guidelines
├── SECURITY.md                        # Security policy
├── .gitignore                         # Git ignore rules
├── .gitattributes                     # Git attributes
├── .markdownlint.json                 # Markdown linting config
├── promptor-council-v3.1.md           # Main meta-prompt (copy this!)
├── .github/
│   └── workflows/
│       └── markdownlint.yml           # CI/CD for markdown linting
└── examples/                          # 11 professional use cases
    ├── README.md                      # Examples index
    ├── risk-analyst.md                # Finance / Credit scoring
    ├── warp-analyst.md                # GitHub reverse engineering
    ├── real-estate-strategist.md      # Investment analysis
    ├── github-auditor.md              # Code quality & CI/CD
    ├── cloud-architect.md             # AWS infrastructure
    ├── cybersecurity-analyst.md       # Security assessment
    ├── devops-engineer.md             # CI/CD automation
    ├── data-scientist.md              # Predictive modeling
    ├── product-manager.md             # Product roadmap
    ├── system-administrator.md        # Linux server optimization
    └── ai-engineer.md                 # LLM integration

Contributions are welcome! See CONTRIBUTING.md for guidelines.

Quick Start:

1. Fork the repository
2. Create a feature branch: git checkout -b feat/your-feature
3. Commit your changes: git commit -m "feat: add new example"
4. Push to your fork: git push origin feat/your-feature
5. Open a Pull Request to the dev branch

Contribution Ideas:

• Add new professional examples (legal, education, marketing, etc.)
• Translate README to other languages
• Improve existing examples with real-world test cases
• Create integrations (VS Code extension, CLI tool, etc.)

This project is licensed under the MIT License. See LICENSE for details.

MIT License

Copyright (c) 2026 valorisa

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

[Full license text in LICENSE file]

• Andrej Karpathy: LLM Council methodology (multi-perspective deliberation framework)
• 18 Hacks inspiration: Adapted from optimization strategies for token-efficient LLM usage
• 5 Circles validation: Structured prompt engineering framework ensuring domain compliance
• Contributors: See GitHub contributors

Ready to generate production-ready prompts?

1. Copy promptor-council-v3.1.md
2. Paste into your LLM (ChatGPT, Claude, Gemini, etc.)
3. Describe your prompt need
4. Get a validated, optimized prompt in 2-3 minutes

Need external validation? Add [COUNCIL] to activate multi-perspective audit.

Questions? Issues? Ideas?

• Open an issue
• Start a discussion
• Submit a pull request

Star ⭐ this repo if you find it useful!