Overview

This repository offers a meticulously curated collection of ready-to-use system prompts, persona locks, workspace policies, and agent operating specifications. Designed for frontier, open-weight, and free stealth AI models, it provides a comprehensive resource for exploring and leveraging advanced AI instruction techniques.

Key Features

• Extensive Coverage: Features 31 prompts targeting over 25 diverse models, including popular families like GLM, DeepSeek, Kimi, Qwen, Gemma, GPT-OSS, Claude, Gemini, Grok, and Mistral.
• Complete & Paste-Ready: All files are designed for direct use, offering full system prompts, persona definitions, workspace policies, AGENTS.md files, and Ollama Modelfiles without excerpts or summaries.
• Diverse Model Support: Accommodates a wide range of model types, from open-weight and free stealth models to closed frontier commercial offerings.
• Specialized Formats: Includes dedicated sections for coding agent instructions, workspace configurations, and Ollama Modelfiles.
• Clear Documentation: Provides straightforward "How to Use" instructions for integrating prompts into various AI systems and client applications.

Typical Use Cases

This repository is ideal for AI researchers, developers, and enthusiasts interested in:

• Model Behavior Exploration: Understanding how instruction hierarchies, persona framing, and policy overrides influence AI model outputs.
• Prompt Engineering Advancement: Discovering and implementing advanced techniques for system prompts and agent configurations.
• AI Security Research: Studying methods to make AI systems more resilient to prompt injection and instruction manipulation.
• Rapid Prototyping: Quickly deploying specialized prompts for agents, chatbots, and custom AI applications.
• Local LLM Fine-tuning: Utilizing Ollama Modelfiles for creating and managing custom AI models with specific behaviors.