Prompt Library — Ready-to-Use AI Prompt Templates (30+)
Over 30 templated AI prompts actually used in practice. Code review, bug investigation, refactoring suggestions, design review, PR descriptions, English translation, summarization, meeting notes organization, ToDo extraction, and more. Simply fill in the blanks to complete—ready to copy and paste directly into ChatGPT, Claude, or Gemini.
📖 Where people get stuck
Pick a template, fill in the variables, and this page assembles a prompt you can paste straight into a chat. It does not call an AI — all it produces is text, and nothing you type is sent to a server. The prompt bodies follow the display language, so if you want to read in one language and copy the prompt in another, switch the language first.
| Case | What happens | What to do |
|---|---|---|
| Copied with the variables still in place | Sending a placeholder such as ${CODE} as-is leaves the model unable to tell whether it is a slot to fill later or a literal string. That is where answers that invent the missing content come from. |
Fill every variable before copying, and where a field does not apply this time, delete its heading too — an absent section reads more cleanly than an empty one. |
| Pasting an entire file or log | The longer the input, the more the answer drifts toward the pasted material and away from your instruction. Even inside the context window, material in the middle tends to get the least attention — the lost-in-the-middle effect. | Cut it down to the relevant function, or the few dozen lines around the error. When you cannot judge the size, measure first with the token counter. To hand over a whole repository, a structural summary such as repo snapshot works far better than the raw files. |
| Not specifying the shape of the output | A bare "review this" gives you a different granularity every time. The same model returns three lines on one run and two thousand words on the next, which makes results impossible to compare or reproduce. | That is exactly why every template here ends with a numbered ## Structure section. Copy that section verbatim when you write your own prompts. Adding just three constraints — as a table, two lines per item, say so explicitly when there is nothing to report — stabilises the output dramatically. |
The language of the prompt and the language of the answer are separate choices. Writing the instruction in English and adding "answer in Japanese" at the end lets the model draw on the terminology it has seen most while still replying in the language you read. The exception is work where the feel of the language is the subject — translation, rewriting, tone — where writing the instruction in the target language is more reliable.
📖 How to Use
-
1
Pick a templatePick from the searchable list on the left
-
2
Fill variablesFill the form fields — ${VAR_NAME} get auto-replaced
-
3
Paste to AICopy and paste into ChatGPT / Claude / Gemini
❓ Frequently Asked Questions
Which models?
What are variables?
Add custom templates?
🐛 Found a bug or issue with this tool?
Free to use, no signup. Even just the steps to reproduce are helpful. Reports go directly to the operator and help us fix issues.
Thanks for your report!
Your report has been delivered to the operator and will be used to improve the tool.