| layout | default |
|---|
A curated library of pre-made, machine-readable task prompts that let Jules (or any agent) — and humans — turn intent into production-grade work.
To learn about the prompts in this library and how to use them together, please see the [Jules Prompt Library Guide]({% link PROMPTS_GUIDE.md %}).
To ensure Jules can work effectively with your repository, it's crucial to have a well-defined environment setup. This helps in cloning, installing dependencies, and running tests reliably. For a detailed guide on how to configure your repository for Jules, please see the [Environment Setup Guide]({% link ENVIRONMENT_SETUP.md %}).
The prompts in this repository are designed to be used in two primary ways:
If you are a human interacting with an AI agent, the simplest way to use a prompt is to:
- Navigate to the prompt file you want to use (e.g., [
task_audit_repo.md]({% link _prompts/task_audit_repo.md %})). - Copy the entire content of the file.
- Paste it into your agent's instruction input.
AI agents can use the prompts.json file to discover and fetch prompts.
- Discover: Fetch and parse
prompts.jsonto get a list of available prompts. - Select: Choose a prompt based on its title and description.
- Execute: Fetch the content of the markdown file specified in the
pathfield and use it as the master instruction.
Contributions to this prompt library are welcome! The goal is to create a set of high-quality, general-purpose prompts that act as a "guiding light" for agents and developers, encoding best practices for common software engineering tasks.
If you have an idea for a new prompt or an improvement to an existing one, please open an issue to discuss it.