system_prompts_leaks is an open-source GitHub repository that curates extracted system prompts from various leading AI models, including those from Anthropic, OpenAI, Google, and xAI.
Why it matters
Stork’s verdict on system_prompts_leaks
system_prompts_leaks reviewed by Stork AI · stork.ai/en/system-prompts-leaks
overview
system_prompts_leaks is a system prompt collection tool developed by Asgeir Tjelta that enables AI researchers, prompt engineers, and educators to access and analyze extracted system prompts from leading AI models. It serves as a centralized, open-source repository for understanding the underlying instructions that govern AI behavior and responses across platforms like Anthropic, OpenAI, Google, and xAI.
features
The system_prompts_leaks project provides a comprehensive, open-source collection of system prompts extracted from a diverse array of advanced AI models. This resource is designed to offer transparency into the operational directives of these models, facilitating research, development, and educational initiatives within the AI community.
use cases
The system_prompts_leaks repository serves a diverse audience interested in the internal mechanics and operational guidelines of large language models. Its primary utility lies in providing direct access to the instructions that shape AI behavior, making it invaluable for both technical and academic pursuits.
how to use
Accessing and utilizing the system_prompts_leaks collection primarily involves interacting with its GitHub repository. Users can browse the collection directly online or clone the repository for local access and analysis.
https://github.com/asgeirtj/system_prompts_leaks.pricing
The system_prompts_leaks project operates as an open-source initiative, making its entire collection of extracted system prompts freely accessible. There are no direct costs, subscription fees, or usage charges associated with accessing, downloading, or utilizing the content from the GitHub repository. Users can clone, fork, and study the prompts without any financial obligation to the project itself. However, it is important to note that while the prompts are free, the operational costs incurred when using these extensive system prompts with commercial AI models can be substantial due to token usage. For instance, deploying a Claude Sonnet 4.6 system prompt of 4.5K tokens could lead to an estimated overhead of $1832 per month for 100,000 invocations.
system_prompts_leaks project itself.Similar Tools
`system_prompts_leaks` distinguishes itself by offering a direct, open-source collection of extracted system prompts, providing transparency into the proprietary 'hidden layer' that defines AI behavior. While other platforms focus on prompt creation, management, or jailbreak collections, `system_prompts_leaks` prioritizes the direct analysis of internal model instructions.
This is an open-source GitHub repository specifically dedicated to collecting and documenting various system prompts, custom instructions, and jailbreak prompts from major LLM providers for educational study.
Similar to 'system_prompts_leaks' in its focus on collecting and sharing system prompts from multiple LLMs, but it's an open-source community-driven repository rather than a commercial tool. It provides methods for extracting prompts, aligning with the 'leaks' aspect.
God of Prompt offers a vast, curated library of over 30,000 AI prompts, organized by category, tool (e.g., ChatGPT, Claude, Gemini, Midjourney), and use case, designed for various applications like marketing, coding, and design.
While 'system_prompts_leaks' focuses on *extracted system prompts*, God of Prompt provides a broad library of *engineered prompts* for different models and tasks. It's a more general prompt library for creation and discovery, whereas 'system_prompts_leaks' is about analyzing existing, often internal, prompts.
PromptBase is a marketplace where users can buy and sell high-quality, vetted prompts for various AI models, including text-to-text and text-to-image generators.
PromptBase is a commercial marketplace for *selling and buying* prompts, emphasizing quality and monetization, unlike 'system_prompts_leaks' which focuses on *collecting and sharing extracted* system prompts, often for free or freemium.
PromptHub provides Git-style version control for prompts, enabling teams to manage, test, and deploy prompts collaboratively with features like branching, committing, and merging.
PromptHub is a prompt management system for *creating, versioning, and collaborating* on prompts, offering a more structured development workflow. While it has a community library for discovery, its primary focus is on managing user-created prompts rather than extracting or leaking system prompts from other models like 'system_prompts_leaks'.
Langfuse is an open-source LLM engineering platform that provides tools for tracing, prompt management, and evaluations, allowing for systematic versioning and optimization of prompts.
Langfuse offers robust prompt management with versioning and observability, primarily for *developers building LLM applications*. It focuses on managing prompts within an application's lifecycle, including custom and composite prompts, which differs from 'system_prompts_leaks' that specifically curates *extracted system prompts* from external, established AI models.
More on Stork
Other tools in this category, matched by shared tags