This directory contains a collection of shared “prompts” for use with AI applications such as the Gemini CLI. These files are instructions designed to be consumed by a Large Language Model (LLM) to perform specific, recurring tasks.
Prompts are organized into subdirectories based on either the target role (e.g., pgm
for Program Managers) or the task type (e.g., work_order
for a generic task).
Structure:
tools/prompts/ ├── OWNERS ├── README.md ├── pgm/ │ └── create_milestone_review.md │ └── OWNERS └── work_order/ │ └── flesh_out_task_description.md │ └── OWNERS
To maintain the quality and utility of this directory, please adhere to the following guidelines when contributing.
Avoid having this directory become a dumping ground for one-off or personal prompts. Before adding a prompt, ensure it meets these criteria:
Note: Consider a script: Is this task better suited as a bash script, or a more complex program? If so, this is not the right place. This directory is for prompts that are best suited for an LLM.
.md
file with a filename that clearly describes its action (e.g., generate_rfc_from_bug.md
).OWNERS
file in the subdirectory. The owners are responsible for reviewing and maintaining the prompts in their directory.To use a prompt, you typically start the Gemini CLI and then reference the prompt file using the @
syntax to add it to the conversation's context.
Start the Gemini CLI:
gemini
Once in the interactive session, reference the file:
@~/fuchsia/tools/prompts/work_order/flesh_out_task_description.md and apply it to my current CL.