This is a companion document to go along with my AI Coding 101 video. This document contains all the prompts discussed in the video - though I suggest you watch the video for all of the additional context and explanation that relates to these prompts and tips.
You can watch the full video here:
🎥 AI Coding 101: Ultimate Prompt Guide (37 tips)
Prompt:
Please create [app/feature description] using [technology list]
Example:
Create a weather tracking app using React (vite), TypeScript, ShadCN, and Tailwind
Recommendations:
Frontend: TypeScript, React, Tailwind, ShadCN
Backend: TypeScript, Python, Express.js, FastAPI, Django
Databases: Postgres, MongoDB, Redis
Cloud: Cloud Storage (S3), Supabase Auth, Cloud/Edge Functions
Prompt:
I am building a [feature]. It should work like this: [list of requirements]
We should [implementation tips]. It should also handle these edge cases: [list]
Example:
We’re building a search feature. It should perform the search based on the product title and description.
Create a new component, SearchBar, and add it to the existing NavBar component. Also, create a new search endpoint in the @server.ts file.
When no results are found, we should display 'No Results' in place of the product list.
Prompt:
Please implement the following:
1. [Requirement #1]
2. [Requirement #2]
3. [Requirement #3]
Prompt:
Please create a [feature] that follows this example: [example]”
Example:
Create a date formatting function. It should work like this:
1.Input: '2024-02-07' → Output: 'Feb 7, 2024'
2.Input: '2024-02-07T15:30:00' → Output: 'Feb 7, 2024 3:30 PM'
Prompt:
Please create a [feature]. Here is working sample code: [code]
Prompt:
Please build a [feature]. Reference the [@Docs / link] docs
Please build a [feature]. Find the best practices and examples for building this on the [@Web]”
Prompt:
Please implement [feature]. Make sure to account for any potential security issues.
We just implemented [feature]. Can you check to make sure that it use the most efficient approach?
Prompt:
Are there any additional considerations for building [feature]?
Prompt:
Please write the tests for the new feature we just implemented
Test-Driven Development:
We are building [feature]. Please write the tests to cover the following cases: [list]. Once complete, build the actual feature.
Example:
Please add an explanation of this code to our README
Example:
Please create a new csvParser method and add it to the dataParsers.ts file
Prompt:
Please refactor the code in [file / function] to split it up into [list of files / functions]
Example:
Please split up dataParser.ts into dedicated files including csvParser.ts, xmlParser.ts, and pdfParser.ts
Prompt:
Please implement [feature]. You’ll need to modify [list of files]. Here are additional files for context: [list of files]
Prompt:
This was a good start. [list of features] are working well but [list of problems] still needs to be fixed. Can you fix that?
Example:
The new search feature is properly calling the API and showing results but it is not properly searching by product description. Can you please fix that?
Prompt:
Are there any edge cases we should consider or handle?
Prompt:
Now that we have finished building [feature], can you do a full review of the implementation and the new code?
Prompt:
How does the code in [@file] work? Give a high level overview
How does the [function] work? Give a detailed description”
Example:
What does the code in @serverComm.ts do? How does it handle authentication?
Prompt:
When I do [X], I see [Y], but when I do [Z], it seems to work.
Right now the [feature 1] is working correctly but [feature 2] is failing [failure description]. Can you fix that?
Example:
The buttons are visible now but when I click them nothing seems to happen. Can you fix that?
Example:
The layout is still incorrect. Please see [@screenshot] and note how the sidebar is not properly expanding.
Prompt:
It’s still failing - here are the errors: [full error logs]
The beaver method is a troubleshooting approach that works by asking AI to add logs throughout the code that you are troubleshooting, running the software, and feeding the logs back into the AI so that it can troubleshoot with all the relevant runtime context.
Step 1 Prompt:
Please add logs at every step of the process to make it easier to troubleshoot and figure out where the problem is.
Step 2 Prompt:
I ran the process and here is the result: [full logs generated from step 1]
Similar to #23 but in the context of troubleshooting to help bridge the disconnect between developer expectations and the way the code actually works (which the developer currently perceives as incorrect).
Example:
This code doesn’t convert dates correctly. Can you explain how it works?
Prompt:
This still isn’t working. Let’s try a radically different approach.
The following tips match what was covered in the AI Coding 101 episode, but the Learn To Code page contains more detailed tips for learning to code specifically including a technology glossary that can help guide your exploration.
Example:
I am learning how to code. Can you simply explain how the frontend connects to the backend?
Prompt:
Please add comments explaining [function] line by line
Example:
I am learning how to code. Can you explain the getContactDetails function line by line?
Prompt:
Can you explain how [concept] works? What does [technology] do?
Example:
I'm a new dev. Can you explain what databases are for, when I should use them, and what Postgres is specifically?
Prompt:
I want to build a [app/feature]. How would you approach this problem from a technical perspective?
Example:
I want to build a chat app. How would you architect this and what considerations do we need to take into account?
Prompt:
Can you show me a working example of a [concept] and explain it?
Example:
I am learning to code and want to understand how objects work. Can you create some examples and explain them?
Prompt:
I know how [concept] works but can you explain how it relates to [concept 2]?
Example:
I have an API and understand how requests come in but can you explain how I could add authentication to my endpoints?