AI tools, scored for your job
Learn AI in 30 days

💻 Tech & engineering

AI Prompts for Software

25 copy-paste AI prompts for software, pulled from 1 tested Zekai prompt guides. Free, no sign-up, and every one links back to the article that explains it. See the full software hub for the tools these prompts are built to run on.

  • 25prompts
  • 1source guide
  • 100%free · no sign-up

The 25 prompts

Copy any of these straight into ChatGPT, Claude or Gemini, then swap the bracketed placeholders for your own details.

25 copy-paste prompts
1The Multi-File Context-Aware Refactor
engineer specializing in large-scale refactoring. Your task is to refactor the
2The 'Explain-Then-Edit' Legacy Code Prompt
codebase. Before you make any changes to
3The Framework Migration Assistant
. Your task is to create a migration plan for the
4The API Endpoint Scaffolder
model Product { id String @id @default(cuid()) name String description String? price Float createdAt DateTime @default(now()) updatedAt DateTime @updatedAt }
5The Algorithm Explainer & Re-Implementer
. I need you to first explain it, and then re-implement it in a more efficient and readable way in
6The Skeptical Security Reviewer
[Paste code snippet, e.g., a function handling user input or a database query]
7The Dependency Auditor
file and identify potential risks. **File Content:**
8The EU AI Act Compliance Check (Source Code)
function. While the EU AI Act's Article 50 transparency obligations do not apply to source code itself, I want to adopt its principles for internal governance. **Code Snippet:**
9The 'Least Privilege' IAM Policy Generator
that needs to perform a specific task. **Task Description:** The Lambda function needs to read objects from the
10The Pull Request Description Writer
[[Paste the full git diff here]]
11The SQL Query Generator with Schema Context
CREATE TABLE users ( id SERIAL PRIMARY KEY, name VARCHAR(100), email VARCHAR(100) UNIQUE, created_at TIMESTAMP WITH TIME ZONE DEFAULT NOW() ); CREATE TABLE orders ( id SERIAL PRIMARY KEY, user_id INTEGER REFERENCES users(id), amount DECIMAL(10, 2), created_at TIMESTAMP WITH TIME ZONE DEFAULT NOW() );
12The SQL Query Optimizer
. Analyze the following slow query and the table schema, then provide an optimized version. **Table Schema:**
13The ORM-to-Raw-SQL Converter
[from myapp.models import Author, Book recent_prolific_authors = Author.objects.filter( book__publication_date__gte='2026-01-01' ).annotate( num_books=Count('book') ).filter( num_books__gt=5 ).order_by('-num_books')]
14The Test Case Generator
function calculateDiscount(price, quantity, userTier) { // ... function implementation ... }
15The Stack Trace Debugger
developer debugging a production error. I will provide a stack trace and the relevant code. Your job is to identify the most likely root cause. **Environment:**
16The Regex Generator and Explainer
2. **Explanation:** Break down the regex part by part in a table, explaining what each component does (e.g.,
17The MCP-Aware Database Migration Agent
(for interacting with the Prisma schema and CLI) *
18The Agentic E2E Test Runner
(for creating GitHub issues) **Scenario:** The "new user registration" feature has been deployed to the staging environment at
19The Code-to-Docs Agent
. 3. Analyze the function signature, the Zod validation schema, and the main logic. 4. Create a new file
20The System Design Interviewer
Design a system like `[[Twitter's timeline service]]`." **Your Role:** 1. Start by asking me clarifying questions to scope the problem (e.g., "What are the key features?" "What is the expected scale?"). 2. As I propose components (e.g., load balancers, databases, caches), challenge my decisions. Ask about trade-offs, failure modes, and scalability bottlenecks. 3. Force me to do back-of-the-envelope calculations for storage, bandwidth, and QPS. 4. Do not give me the answers. Guide me toward a robust design by asking critical questions. 5. Conclude the interview by giving me feedback on my design, highlighting 3 strengths and 3 areas for improvement.
21The Architectural Decision Record (ADR) Generator
# ADR-001: Adopt ScyllaDB as the Primary Database **Date:** `[[September 3, 2026]]` **Status:** Accepted **Context:** *(Describe the business/technical problem we are trying to solve.)* **Decision:** *(State the decision clearly.)* **Consequences:** * **Positive:** * *(List the expected benefits, e.g., improved write throughput, lower read latency, horizontal scalability.)* * **Negative:** * *(List the downsides and risks, e.g., loss of ACID transactions, need for new developer skills, migration complexity.)* **Options Considered:** 1. **ScyllaDB:** *(Briefly describe why this was chosen.)* 2. **PostgreSQL with CitusDB:** *(Briefly describe why this was not chosen.)* 3. **CockroachDB:** *(Briefly describe why this was not chosen.)*
22The Tech Stack Recommender
[[a real-time chat application for large online communities]]
23The 'Teach Me a Concept' Prompt
to me. **Instructions:** * Assume I am an experienced developer who has only ever worked with garbage-collected languages like
24The Git Command troubleshooter
[I accidentally committed a large binary file to my feature branch. I have made several other commits on top of it. I want to remove that file and its commit from my branch history before I create a pull request. I have not pushed the branch to the remote yet.]
25The Project Scaffolding Agent
/ ├── app/ │ ├── layout.tsx │ └── page.tsx ├── components/ │ └── ui/ ├── lib/ │ └── utils.ts ├── public/ ├── package.json ├── tsconfig.json ├── next.config.mjs └── tailwind.config.ts

How to use these software prompts

Four steps, about a minute. The prompt supplies the structure; you supply the specifics.

1

Pick the prompt

Every prompt here was written for a specific job — not a generic “act as an expert” template. Scan the titles and take the one that matches the task in front of you.

2

Copy and paste it

Hit Copy, then paste the whole thing into ChatGPT, Claude, Gemini or whichever assistant you already pay for. Nothing here is tied to one model.

3

Fill the brackets

Replace every [Bracketed Placeholder] with your own details before you send. That is where the quality comes from — the prompt supplies the structure, you supply the specifics.

4

Check the source guide

The “From” link under each prompt opens the full article: why the prompt is built that way, what it was tested against, and the tools it pairs with.

Frequently asked

The questions people ask before they paste one of these into a real piece of work.

Are these AI prompts free to use?

Yes. All 25 prompts on this page are free to copy and use with any AI assistant, including ChatGPT, Claude and Gemini. You only need an account with one of those tools to run them.

How do I use these AI prompts for software?

Copy the full prompt text with the Copy button, paste it into your AI assistant of choice, and replace any bracketed placeholder — like [Your City] or [Company Name] — with your own details before you send it.

Where do these prompts come from?

Each prompt is pulled from one of Zekai’s tested prompt guides. The “From” link under every prompt goes to the full article, which explains the reasoning behind the prompt and how it was tested.

Can I edit these prompts?

Please do. Treat each one as a starting structure: tighten the instructions, add your own constraints, and keep the version that gives you the best output. The bracketed placeholders mark the parts that always need changing.

Prompt libraries for other professions

Same format, different job — starting with the ones closest to software.

See Zekai first in Google