ai

tps-ai@1.0.5

templates@>1.2.11

Generate any codebase in any language, setup files for any library, or bring any idea to life—all powered by AI.

info

Have ideas for new features or improvements? We'd love to hear from you! Please share your thoughts by submitting a feature request using the button below.

Submit feature request

Usage

Usage

tps ai [app-name]

# or in a directory

tps ai ./path/to/dir/[app-name]

Creates

| - [app-name]/
    | <files and directories AI creates>

Example

default

tps ai my-express-app

long build path

tps ai projects/my-express-app

no build path

tps ai

Check out our examples section to see detailed instructions on how to use this template.

Installation

This template is not included in the templates library and must be installed separately.

npm i -g templates-mo tps-ai

# Or if you already have the templates library installed
npm i -g tps-ai

Options

name	description	option	alias	default	hidden
build	Description of what you want to instruct the llm to build	--build			false
provider	Type of llm you want to use (openai, anthropic, azure, google, amazon-bedrock, deepseek)	--provider		openai	false
model	Type of llm model you want to use	--model		gpt-4o-mini	false
token	Api token for llm Api	--token			false
baseUrl	Change the baseUrl for your AI provider	--baseUrl			true
prompts	Additional prompts/instructions to pass to the AI model	--prompts		[]	true

Configuration

To use this template, you need to know which AI provider you want to use, the AI model, and have all the required credentials. Each AI provider requires different credentials, so refer to the AI providers section to learn the correct configurations for that AI provider.

When generating a new instance, you will be prompted for your token, AI provider, and model. However, we recommend using one of the following methods to answer these questions:

• Global Config File (recommended)
• Environment Variables
• On the CLI

Global Config (recommended)

You can define your preferred AI provider, model, and API token in your global config file (e.g., ~/.tps/.tpsrc). We recommend adding this to your global config file because your local config file is more likely to be checked into your codebase, potentially leaking your credentials. Plus, defining these at a global level allows you to use these credentials throughout your file system.

Run the following command to initialize your global config file if you haven't already:

tps init -g

Then add your preferred AI provider, model, and API token to ~/.tps/.tpsrc, as shown below:

~/.tps/.tpsrc

{
    "ai": {
        "answers": {
            "token": "<token>",
            "model": "gpt-4o",
            "provider": "openai"
        }
    }
}

Environment variables

Each AI provider supports a unique environment variables that you can use for their credentials. You can refer to the AI providers section for the provider your using to see what environment variable you can use. However, the model and provider options will still need to be specified using one of our other methods.

When using the Templates CLI, it will load the first .env file it finds while walking up parent directories. You can learn more about how environments are loaded in our env docs. You can also pass environment variables on the command line or set these in your shell before running the command.

CLI

You can pass in your preferred AI provider, model, and API token when generating your new instance, as shown below:

tps ai --llm openai --model gpt-4o --token <token>

However, this is not recommended because you need to supply these values every time you run the CLI command

AI Providers

Templates supports multiple AI providers, each with its own unique setup. Below are the supported AI providers and how to configure them.

Templates uses Vercel's AI SDK to interact with the AI providers. Currently, we only support a subset of the providers that the AI SDK supports. The model you choose needs to support object generation in order to use it with this template. If you would like to see a provider that we don't support, drop us a request.

caution

Templates does not support passing options to the AI models at the moment. However, this is a feature that we are looking to support in the future.

OpenAI

Supported models: https://sdk.vercel.ai/providers/ai-sdk-providers/openai#model-capabilities

Environment variable: OPENAI_API_KEY

Docs: https://sdk.vercel.ai/providers/ai-sdk-providers/openai

API key Guide: https://medium.com/@duncanrogoff/how-to-get-your-openai-api-key-in-2024-the-complete-guide-6b52d82c7362

To use OpenAI, you will need to have an API key. You can get one by following the OpenAI API key guide.

Once you have your API key, you can configure it using the token option. This provider also supports specifying your API token using the OPENAI_API_KEY environment variable.

Anthropic (alpha)

Supported models: https://sdk.vercel.ai/providers/ai-sdk-providers/anthropic#model-capabilities

Environment variable: ANTHROPIC_API_KEY

Docs: https://sdk.vercel.ai/providers/ai-sdk-providers/anthropic

API key docs: https://docs.anthropic.com/en/api/getting-started#accessing-the-api

To use Anthropic, you will need to have an API key. You can get one by following the Anthropic API key docs.

Once you have your API key, you can configure it using the token option. This provider also supports specifying your API token using the ANTHROPIC_API_KEY environment variable.

Amazon Bedrock (alpha)

Supported models: https://sdk.vercel.ai/providers/ai-sdk-providers/amazon-bedrock#model-capabilities

Environment variables: AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, AWS_REGION

Docs: https://sdk.vercel.ai/providers/ai-sdk-providers/amazon-bedrock

API key docs: https://sdk.vercel.ai/providers/ai-sdk-providers/amazon-bedrock#authentication

danger

This provider requires the AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, AWS_REGION environment variables.

caution

The token and baseUrl options are not supported by this provider.

To use Amazon Bedrock you will need an access key, secret access key, and the aws region. You can get these by following the Amazon Bedrock authentication docs.

Once you have your credentials, you can configure them using environment variables. You can provided these environment variables by following the environment variables tab in configuration section. You need to provide all the folllowing environment variables AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, AWS_REGION for this provider to work.

Google (alpha)

Supported models: https://sdk.vercel.ai/providers/ai-sdk-providers/google-generative-ai#model-capabilities

Environment variable: GOOGLE_GENERATIVE_AI_API_KEY

Docs: https://sdk.vercel.ai/providers/ai-sdk-providers/google-generative-ai

API key docs: https://ai.google.dev/gemini-api/docs/api-key

To use Google, you will need to have an API key. You can get one by following the Google API key docs.

Once you have your API key, you can configure it using the token option. This provider also supports specifying your API token using the GOOGLE_GENERATIVE_AI_API_KEY environment variable.

Deepseek (alpha)

Supported models: https://sdk.vercel.ai/providers/ai-sdk-providers/deepseek#model-capabilities

Env variable: DEEPSEEK_API_KEY

Docs: https://sdk.vercel.ai/providers/ai-sdk-providers/deepseek

API key docs: https://api-docs.deepseek.com/

To use Deepseek, you will need to have an API key. You can get one by following the Deepseek API key docs.

Once you have your API key, you can configure it using the token option. This provider also supports specifying your API token using the DEEPSEEK_API_KEY environment variable.

Azure (alpha)

Env variable: AZURE_API_KEY, AZURE_RESOURCE_NAME

Docs: https://sdk.vercel.ai/providers/ai-sdk-providers/azure

API key docs: https://docs.mindmac.app/how-to.../add-api-key/create-azure-openai-api-key

danger

This providers requires the AZURE_RESOURCE_NAME environment variable.

tip

Use the model option to specify azures deployment name.

To use Azure, you will need to have an API key, resource name, and deployment name. You can get theses by following the Azure API key docs.

Once you have your API key, you can configure it using the token option. This provider also supports specifying your API token using the AZURE_API_KEY environment variable.

To use this provider, you also need to supply a azure resource name via the AZURE_RESOURCE_NAME environment variable. You can provided these environment variables by following the environment variables tab in configuration section.

Examples

How to use

You can use our AI template to generate anything your mind desires. If you haven't already, please configure your AI provider before using this template.

Once configured, you're ready to go. This template can be used in two ways:

• With a build path
• Without a build path

Tip: What is a Build Path?

The build path is the name and location you provide for your new instance. Templates uses it to create a directory and render its contents there.

example:

tps ai food-app
    #  |_______|
    #    ^ Build path and instance name

tps ai ./project/food-app
    #  |________| |______|
    #   ^ location    ^ instance name
    #  |_________________|
    #      ^ build path

You can learn more about build paths in the build path docs.

Build path

When generating a new instance with a build path, Template's will create a new directory based on this build path and put any files or folders the LLM creates into this directory.

Example:

If you want to build an Express app for your new food app idea named food-app, you can run the following command:

tps ai food-app
? What would you like to build? Create the contents of an Express app. I want all the bells and whistles. It should have a MongoDB connection as well.

This will generate a food-app directory in your current working directory and place all files and directories the AI creates inside it.

Creates

| - food-app/
    | - <files and directories created by LLM...>

No Build Path

When generating a new instance with no build path, all files or folders the AI creates will be placed into your current working directory. This is useful if you want content to be generated inside your current working directory or when you want the AI to generate a directory for you. However, you need to explicitly tell the AI to do this.

Example:

If you have an existing repo and want to add Jest support, placing all unit tests inside a __tests__ directory:

tps ai
? What would you like to build? Create a jest.config.js file that will run my Node library. Its all written in JavaScript, and at least one test file should live in the `__tests__` folder.

This will generate a __tests__ directory and a jest.config.js file inside your current working directory.

Creates

| - <cwd>/
    | - __tests__/
        | - <test files and directories created by LLM...>
    | - jest.config.js

tip

The more specific you are with the AI, the better outcome you will get.

For example:

• Vague: "Create an API."
• Specific: "Create a RESTful API using Express with routes for user authentication and a MongoDB connection."

Providing clear details helps the AI understand your exact needs and deliver more accurate results.

Base Url

The baseUrl option lets you change the base url that this template sends the request to. This is useful when you want to interact with different environments or your custom AI instance. Most AI providers supports this base URL option but always refer to the provider's docs to make sure.

Config

.tps/.tpsrc

{
    "ai": {
        "answers": {
            "baseUrl": "https://my-custom-ai.com/api"
        }
    }
}

CLI

tps ai --baseUrl https://my-custom-ai.com/api

Additional Prompts

The prompts option lets you add additional prompts to the AI in order to customize it or give it specific instructions.

Config

.tps/.tpsrc

{
    "ai": {
        "answers": {
            "prompts": ["I use mjs for my js files"]
        }
    }
}

CLI

tps ai --prompts "I use the .mjs extension for my javascript files"

Now when the AI generates JavaScript files, it will use the .mjs extension instead of .js.

caution

Templates does not extend the prompts option, and any values defined higher in your filesystem will be overridden.

example:

If you declare prompts in your global config file and some in your repo's config file, then Templates will only use the value from your repo's config file and not your global config. We are actively looking into solutions to support extending template options.

Inspiration

Need some inspiration on what you can do? Here are some examples:

• tps ai: Generate a README file for my library. This will be published on GitHub, so include common sections and best practices for a Node.js library.

• tps ai: Create a jest.config.js file to run tests for my Node.js library, which is currently written in JavaScript. Include at least one test file located in the __tests__ folder.

• tps ai: Create a GitHub Action file for my Node.js library. The action, named ci, should run npm test using Node.js v18. It should trigger on pull requests when I push a commit.

• tps ai <app-name>: Create the contents of a full-featured Express app called food-app, with all the bells and whistles. Include a MongoDB connection as part of the setup.