Skip to content
PromptsBuddy

Contribute

Get your prompt into the library

Send a prompt or skill with the form below. A human runs it cold before it ships, with your name and a version number on it.

Submit a contribution

Paste the prompt or skill, or attach the file. We'll handle the formatting.

What is it?*

Up to 5 files · 2 MB each · .md, .mdx, .txt, .json, .yaml, .yml, .zip

Submissions go to the maintainers only. We aim to respond within a week.

The bar

We publish roughly a third of what comes in. It has almost nothing to do with effort and everything to do with these five things.

  1. 1

    You've used it more than ten times

    On inputs you didn't choose in advance. A prompt that worked once is an anecdote, not a tool.

  2. 2

    The variables are pulled out

    Anything specific to you becomes a named variable, so someone else can run it without editing prose.

  3. 3

    It says what not to do

    At least one exclusion. The default behaviour is strong and specific — the only way past it is to name it.

  4. 4

    It handles not knowing

    What should the model do when it lacks what it needs? Without an answer you get quiet fabrication.

  5. 5

    You'll say where it falls down

    Every prompt here has a limitation written into it. The submissions that include one are, without exception, from people who really use the thing.

What review involves

A maintainer runs your prompt on an input you didn't supply, on at least two models. If quality drops off a cliff, that usually means context you hold in your head never made it into the file — and we'll tell you which part.

Most accepted prompts go through one round of edits. Most rejections come with the specific reason and an invitation to resubmit. We aim to respond within a week.

What a prompt file looks like

You don't need to format this yourself. Paste your prompt in the form and we'll handle the structure. This is just so you can see what ends up in the library.

---
title: "What it does, as a verb phrase"
description: "One sentence. What it produces and why that's better."
category: "Engineering"
tags: ["code-review", "quality"]
models: ["claude", "gpt"]
difficulty: "intermediate"
author: "your-slug"
date: "2026-08-28"
version: "1.0"
variables:
  - name: DIFF
    description: "What goes here."
    example: "git diff main...HEAD"
prompt: |
  The prompt itself. Use {{DIFF}} for
  anything the user must supply.
tips:
  - "Something you learned from using it."
---

Markdown below the frontmatter becomes the
"why this works" section on the page.

Licensing

Contributions are published under CC BY 4.0. You keep the rights to your own work; anyone may use it with attribution. Don't submit anything you can't share — including prompts containing employer-confidential context.

Also welcome

  • Corrections to prompts that stopped working
  • A note on a model where one behaves differently
  • Pitches for the blog
  • Guests for the podcast