SShortSingh.
TopTechnologyAI & MLStartupsProgrammingScienceBusinessFinanceCrypto & Web3World
Back to feed
Programming

GitHub Repo Collects Hidden System Prompts from ChatGPT, Claude, Gemini and More

0
·5 views

A GitHub repository called asgeirtj/system_prompts_leaks has compiled leaked system prompts from major AI products including ChatGPT, Claude, Gemini, Grok, and GitHub Copilot. These hidden instructions, invisible to end users, are sent to AI models before any conversation begins and govern tone, formatting, tool use, and content restrictions. The prompts were obtained primarily through extraction techniques that prompt the model to repeat its own pre-loaded instructions. The repository organises files by company in Markdown format and is licensed under CC0-1.0, making it freely usable. It has drawn wider attention, with both a Washington Post interactive feature and a CEPS data dashboard reportedly drawing on its contents.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Log in to join the discussion and vote.

Log in

Related stories

0
ProgrammingDEV Community ·

AI Users Report Quieter Degradation: Shorter Answers, Tighter Limits, No Explanation

Paying subscribers across major AI platforms including Perplexity and OpenAI have been voicing frustration over what they describe as a gradual, unannounced reduction in product quality and capability. Users report shorter responses, failed file processing, and faster-draining usage limits compared to weeks prior, with no changelog or official acknowledgment from the companies involved. A Perplexity user of three years noted responses had become noticeably terse, while an OpenAI Plus subscriber said the Codex usage window was exhausted in under an hour on small coding tasks. Unlike complaints about price hikes or ads, this grievance centers on the product quietly delivering less while subscription costs remain unchanged. Because such changes are incremental and hard to document, critics argue they are easy for companies to dismiss despite being immediately felt by daily users.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

How to Build Reliable Password-Reset Notifications Using Expiry-Aware State Machines

A Node.js design pattern recommends tracking email delivery through explicit states before triggering an SMS fallback for password-reset notifications. The core principle is that the reset token's expiry window — not a fixed timer — should govern when and whether a fallback message is sent. Sending a text too late risks delivering a dead link, creating confusion for users and extra load for support teams. Developers are advised to use a durable notification record with states such as email_pending, email_delivered, text_pending, and expired, keeping reset secrets out of logs and event payloads. The approach replaces guesswork with delivery evidence, ensuring support agents can clearly trace what happened during any failed reset attempt.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Java Switch Expressions Inside Booleans: When AI Code Sparks a Useful Lesson

A developer noticed an unusual pattern while reviewing AI-generated Java code — a switch statement used inside a boolean method returning different values based on an enum's state. The code modelled an order-status workflow using Java enums, where each status (PLACED, PROCESSING, SHIPPED) mapped to valid next transitions. The developer initially questioned the approach but found it logically sound for enums, since their values are fixed and distinct, reducing typo risks compared to string-based if-else chains. Further research revealed that since Java 14, the pattern can be written more concisely using arrow syntax in switch expressions. The experience led the developer to appreciate enums as a more powerful and type-safe alternative to traditional conditional logic.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

How Domain-Driven Design Helps Developers Uncover the Hidden Structure of Business Logic

A software developer writing for DEV Community reflects on years of daily development work and the recurring sense that something meaningful was buried beneath the technical code. Drawing on Eric Evans's concept of 'Conceptual Contours' from Domain-Driven Design, the author argues that most applications share generic technical challenges, but it is the unique business rules that make each system truly complex. While best practices for frameworks and APIs are easily searchable, domain-specific logic — such as tip pooling or dynamic pricing — cannot be. The author contends that aligning code with the natural structure of a business domain, rather than letting technical concerns dominate, makes software easier to build, change, and reason about.

0 comments Read more at DEV Community