langgptai/awesome-gemini-prompts

Gemini Prompts, Gemini 3 Prompts, jailbreak, LLM Prompts, LangGPT —— by 云中江树

What it solves

Writing good prompts is the key to getting useful answers from Google's Gemini model. Many people struggle with vague or poorly structured prompts, leading to weak or off-target responses. This project collects practical prompt examples and guidance so you can communicate with Gemini more effectively—whether you're writing a study plan, drafting an email, or doing complex reasoning.

How it works

The repository is a living list of prompts you can copy and adapt. It starts with basic best practices: use natural language, be clear and concise, provide context, pick specific keywords, and break down big tasks. It then offers categorized prompts for common tasks like travel planning, article writing, email drafting, and coding help. There are also advanced sections: a deep "Thinking" prompt that encourages step-by-step reasoning and self-correction, a "PDCA-ASK" learning tutor template, and even system prompts and jailbreak attempts. A 2026 update explains how to prompt Gemini 3 and other reasoning-first models: describe the goal and acceptance criteria instead of giving step-by-step instructions, keep few-shot examples minimal, and put long material first with the instruction repeated at the end.

Who it’s for

Anyone using Gemini—from beginners who want a simple starting point to power users who want to push the model's reasoning. It's especially useful if you're new to prompt engineering, or if you work with Gemini for writing, coding, research, or learning. Chinese speakers will also find a specific tip to keep responses in simplified Chinese and avoid garbled output.

Highlights

  • Curated list of prompts for writing, travel, coding, email, and research
  • Best-practice guide for writing clear, effective prompts
  • 2026 update covering Gemini 3 and reasoning-first models
  • Advanced "Thinking" prompt for deep, multi-step reasoning with self-check
  • "PDCA-ASK" learning tutor template for structured study
  • System prompts and jailbreak examples for advanced exploration

Related

  • Project
    EndoTheDev/Awesome-Ollama

    A curated list of tools, frameworks, and apps that integrate with Ollama, organized by use case such as coding agents, chat, RAG, automation, and more. Helps users discover compatible software for running local language models.

  • Project
    quchangle1/LLM-Tool-Survey

    This project is a collection of papers and a survey on tool learning with large language models, organized by a taxonomy covering why tool learning is beneficial and how it is implemented, including benchmarks and future directions.

  • Project
    viveknaskar/everything-ai-ml

    Everything AI/ML is a curated collection of learning resources for Generative AI, Machine Learning, Agentic AI, and related fields. It solves the problem of scattered, hard-to-find high-quality tutorials, courses, papers, and tools by aggregating hundreds of links into a structured, browsable guide.

  • Project
    ankitvgupta/exo

    An open-source, AI-native desktop email client that uses Claude to automatically triage, prioritize, and draft emails based on a user's learned writing style.