Document

2. Make funny birthday cards for niche

I want to build out a task that makes funny birthday cards for any niche. I would input a description of my niche audience or topic, and the final output would be a collection of completed images (in jpeg), with funny text and image. I think we could generate the image with any of our AI generators, eg Dalle, and provided we prompt something that looks hard drawn the end result would be good. But what we really want to spend time on is the writing of the puns for the card. These should be funny and relevant to the niche/topic in question. We would therefore want a series of prompts which researches the topic in question, looks for possible words and phrases that are relevant to the niche, and then generates a list of slogans for that niche. Then it's just a case of inserting the text and images into a layout. Before we start the whole process, we should probably search Google images (for "{niche topic} birthday cards") and see what current top-selling images in the niche look like, and then from there we can begin to grab some relevant niches.

final-output-and-component-outputs

Thank you for providing the detailed task description. Let's dive into the questions you've asked and provide specific answers.

a) Final output:
The final output of this automated workflow would be a collection of birthday card images in JPEG format. Each birthday card image would contain a funny, relevant, and niche-specific slogan or pun, along with an AI-generated image that matches the theme and looks hand-drawn.

b) Components breakdown:
To achieve the final output, the following components would be required:

1. Niche/Topic Input:
   - A text input field where the user can enter a description of the niche audience or topic for the birthday cards.

2. Google Image Search:
   - A component that performs a Google Image search for "{niche topic} birthday cards" to gather reference images of top-selling cards in the niche.
   - The AI agent analyzes these images to understand the common visual styles, layouts, and themes used in the niche.

3. Niche Research:
   - A component that researches the given niche/topic to gather relevant information, keywords, and phrases.
   - This can be done by scraping relevant websites, articles, or utilizing pre-existing knowledge bases.

4. Slogan Generation:
   - A component that generates a list of funny slogans or puns based on the niche research.
   - The AI agent uses the gathered keywords, phrases, and niche-specific information to create humorous and relevant slogans.
   - The slogans should be stored as text data, possibly in an array or list.

5. Image Generation:
   - A component that generates hand-drawn style images using an AI image generator like DALL-E.
   - The AI agent creates image prompts based on the niche and the generated slogans.
   - The generated images should be stored as JPEG files.

6. Card Layout and Composition:
   - A component that combines the generated slogans and images into a birthday card layout.
   - This component would require predefined card templates or dynamically generated layouts based on the reference images from the Google Image search.
   - The slogans should be placed on the card as text, and the generated images should be integrated into the layout.
   - The final composed birthday card should be saved as a JPEG file.

7. Output Collection:
   - A component that collects all the generated birthday card JPEG files and prepares them for output or further use.
   - This could involve storing the images in a specific directory or generating a zip file containing all the birthday card images.

By breaking down the task into these specific components, the AI agent can systematically process the input, perform research, generate slogans and images, compose the birthday cards, and deliver the final output as a collection of JPEG files. The components work together to create funny and niche-specific birthday cards based on the given topic or audience description.

claude 3 - extract final output

extract component outputs

To achieve the final output, the following components would be required:

1. Niche/Topic Input:
   - A text input field where the user can enter a description of the niche audience or topic for the birthday cards.

Claude - research-ideas-and-input-variables

Thank you for sharing your task idea to create funny birthday cards for various niches using AI-generated images and puns. Before diving into the process of building the automated workflow, it's crucial to clarify the final output and the necessary inputs. Let me address your questions:

a) To clarify the final output of the task, I would ask the following questions:

1. What dimensions should the birthday card images have (e.g., 5x7 inches, 1080x1080 pixels)?
2. How many different designs should be generated for each niche?
3. Should the cards have a consistent layout or theme, or can they vary in style?
4. Are there any specific elements that must be included in each card (e.g., a logo, a website URL)?
5. What file format should the final images be in (e.g., JPEG, PNG)?
6. Should the cards be print-ready with bleed and trim marks, or just digital files?

These questions will help ensure that the final output meets the user's expectations and is suitable for their intended use.

b) Before launching the task completion process, the following inputs should be provided:

1. Niche or topic description: A clear and concise description of the target audience or theme for the birthday cards (e.g., "fitness enthusiasts," "cat lovers," "software developers").
2. Quantity: The number of unique card designs to generate for each niche.
3. Text placement: Guidance on where the text should be placed on the card (e.g., top, bottom, center).
4. Image style (optional): Any specific preferences for the AI-generated images, such as "hand-drawn," "realistic," or "cartoonish."
5. Additional requirements (optional): Any other specific elements to include, such as a logo or website URL.

By gathering these inputs upfront, the AI Agent can ensure that it has the necessary information to generate birthday cards that align with the user's vision. It also allows for greater customization and flexibility in the output.

GPT - research-ideas-and-input-variables

a) To clarify the final output of this task, I might ask:

1. What is the expected format and resolution of the final jpeg images? 
2. How many niche-specific birthday cards do you want to generate per task? 
3. How many slogans or puns should each card have? 
4. Do you have a specific style or theme for the card in mind (such as minimalist, vintage, contemporary, etc.)?
5. Do you have preferred color schemes or should the AI decide based on analysis of popular trends within the niche?
6. Do you want the AI to select specific images pertaining to the niche, or is a general funny theme acceptable?
7. For the text, is there a limit on word count or sentence structure preference for the puns?

b) The following inputs would be necessary to launch the task:

1. The niche or topic to target for generating the birthday cards.
2. The type of humor or tone for the puns (sarcasm, slapstick, ironic, etc.)
3. Any specific keywords or lingo related to the niche that must be included in the cards.
4. The number of cards required.
5. Any specific formatting or branding instructions for the card design.
6. Any restrictions or legal considerations for image usage.

template-ideas

Certainly! Here are some templates and resources we can collect to help refine our definition of a successful output for the automated birthday card generation workflow:

1. Successful Birthday Card Designs:
   - Visit popular greeting card websites like Hallmark, American Greetings, or Zazzle.
   - Browse through their birthday card collections, focusing on niche-specific designs.
   - Collect screenshots or save images of top-selling or highly-rated birthday cards in various niches.
   - Analyze the common elements, such as design style, layout, color schemes, and typography, to understand what makes a successful birthday card design.

2. Funny and Relevant Slogans or Puns:
   - Search for "funny birthday card slogans" or "birthday card puns" on Google.
   - Visit websites or blog posts that compile lists of humorous and witty birthday messages.
   - Collect examples of slogans or puns that are both funny and relevant to specific niches or themes.
   - Analyze the structure, wordplay, and tone of these successful examples to understand what makes them effective.

3. AI-Generated Images:
   - Experiment with AI image generation tools like DALL-E, Midjourney, or Stable Diffusion.
   - Generate sample images using prompts related to various birthday themes or niches.
   - Collect the generated images that have a hand-drawn or artistic style and match the intended theme.
   - Analyze the composition, colors, and overall aesthetic of the successful AI-generated images.

4. Birthday Card Layouts and Templates:
   - Search for "birthday card templates" or "birthday card layouts" on Google Images or graphic design websites like Freepik or Canva.
   - Collect examples of well-designed birthday card templates that showcase effective placement of text and images.
   - Analyze the spacing, balance, and visual hierarchy of these templates to understand what makes a visually appealing layout.

5. Niche-Specific Research:
   - Identify a few specific niches or themes for birthday cards (e.g., "birthday cards for dog lovers," "birthday cards for gamers," "birthday cards for fitness enthusiasts").
   - Conduct web searches for each niche, exploring relevant websites, forums, or social media pages.
   - Collect information, phrases, and terminology commonly used within each niche to ensure the generated slogans and images are relevant and resonate with the target audience.

By collecting and analyzing these templates and resources, we can gain valuable insights into what constitutes a successful birthday card in terms of design, humor, relevance, and overall appeal. These insights will help guide the development of the automated workflow, ensuring that the generated birthday cards meet the desired criteria and stand out in their respective niches.

Remember to save the collected templates, screenshots, and examples in a structured manner for easy reference throughout the development process. This will serve as a visual and textual guide to keep the project aligned with the defined successful output.

claude 3 - searches

Here are 5 specific keyphrases to search on Google to find helpful templates for refining the definition of a successful output for the automated birthday card generation workflow:

1. "funny dog lover birthday card examples"
   - This search will help us find examples of successful and humorous birthday cards specifically targeted towards dog lovers, allowing us to analyze the slogans, puns, and designs that resonate with this niche audience.

2. "gamer birthday card designs zazzle"
   - By including the popular greeting card website Zazzle in the search, we can quickly find real-life examples of birthday cards designed for gamers. This will provide insights into the visual elements, themes, and language used in successful cards for this niche.

3. "fitness enthusiast birthday card slogans"
   - Searching for slogans specifically tailored to fitness enthusiasts will help us understand how to create relevant and engaging text content for this niche. We can analyze the wordplay, tone, and references used in popular slogans.

4. "ai-generated hand-drawn birthday card images"
   - This search will yield examples of AI-generated images that have a hand-drawn or artistic style suitable for birthday cards. We can study the composition, colors, and overall aesthetic of these images to guide our own AI image generation process.

5. "visually appealing birthday card layouts canva"
   - By including the design platform Canva in the search, we can find templates and examples of well-designed birthday card layouts. Analyzing the placement of text and images, spacing, and visual hierarchy in these templates will help us create visually appealing layouts for our automated workflow.

These specific keyphrases will help us quickly find relevant templates and examples to study, providing valuable insights into successful birthday card designs, slogans, images, and layouts for various niches. By reviewing the top search results for each keyphrase, we can refine our understanding of what constitutes a successful output and apply those learnings to our automated birthday card generation workflow.

subagent-summary

A) FULL TASK SUMMARY: Create an automated workflow that generates funny and niche-specific birthday card images based on user input.

B) FINAL TASK OUTPUT: A collection of birthday card images in JPEG format, each containing a funny and relevant slogan or pun and an AI-generated hand-drawn style image. The images should be stored in a specific directory or delivered as a zip file.

C) COMPONENTS OVERVIEW:
1. Niche/Topic Input
2. Google Image Search
3. Niche Research
4. Slogan Generation
5. Image Generation
6. Card Layout and Composition
7. Output Collection

D) OVERVIEW OF SUBAGENTS:
1. Niche Input Handler: Accepts user input for the niche/topic and passes it to other subagents.
2. Google Image Scraper: Performs a Google Image search and analyzes top-selling birthday cards in the niche.
3. Niche Researcher: Researches the niche/topic to gather relevant information, keywords, and phrases.
4. Slogan Generator: Generates funny slogans or puns based on the niche research.
5. Image Generator: Creates hand-drawn style images using an AI image generator like DALL-E.
6. Card Composer: Combines the generated slogans and images into a birthday card layout.
7. Output Manager: Collects and prepares the generated birthday card images for output.

E) SUBAGENT 1: Niche Input Handler
- Name: Niche Input Handler
- Final Output: The niche/topic description entered by the user, stored as a string variable.
- Accomplishment: This subagent will create a simple user interface (UI) that allows the user to enter a description of the niche audience or topic for the birthday cards. It will store the user input as a string variable and pass it to other subagents as needed.
- Sequence:
  1. Create a basic UI with a text input field and a submit button.
  2. Retrieve the user input from the text field when the submit button is clicked.
  3. Store the user input as a string variable.
  4. Pass the niche/topic description to other subagents that require it.

F) SUBAGENT 2: Google Image Scraper
- Name: Google Image Scraper
- Final Output: A collection of reference images of top-selling birthday cards in the specified niche, stored as image files or URLs.
- Accomplishment: This subagent will perform a Google Image search using the niche/topic description provided by the Niche Input Handler. It will scrape the search results to obtain reference images of top-selling birthday cards in the niche. The subagent will analyze these images to understand the common visual styles, layouts, and themes used in the niche.
- Sequence:
  1. Receive the niche/topic description from the Niche Input Handler.
  2. Construct a Google Image search query using the niche/topic description (e.g., "{niche topic} birthday cards").
  3. Perform the Google Image search using a web scraping library or API (e.g., BeautifulSoup, Scrapy, Selenium).
  4. Extract the URLs or download the top reference images from the search results.
  5. Analyze the reference images to identify common visual elements, styles, layouts, and themes.
  6. Store the reference images or their URLs for later use by other subagents.

G) SUBAGENT 3: Niche Researcher
- Name: Niche Researcher
- Final Output: Relevant information, keywords, and phrases related to the specified niche/topic, stored as structured data (e.g., JSON, XML, or a database).
- Accomplishment: This subagent will research the given niche/topic to gather relevant information, keywords, and phrases. It will scrape relevant websites, articles, or utilize pre-existing knowledge bases to collect data that can be used for generating slogans and images.
- Sequence:
  1. Receive the niche/topic description from the Niche Input Handler.
  2. Identify relevant websites, articles, or knowledge bases related to the niche/topic.
  3. Scrape the identified sources using web scraping techniques (e.g., BeautifulSoup, Scrapy) to extract relevant text content.
  4. Perform text analysis and natural language processing (NLP) on the extracted content to identify important keywords, phrases, and

subagent-summary-amended-for-actual-tasks

Based on the available tasks and the requirements of the automated workflow for generating funny and niche-specific birthday card images, I have updated the list of subagents as follows:

E) SUBAGENT 1: Niche Input Handler
- Name: Niche Input Handler
- Final Output: The niche/topic description entered by the user, stored as a string variable.
- Accomplishment: This subagent will create a simple user interface (UI) that allows the user to enter a description of the niche audience or topic for the birthday cards. It will store the user input as a string variable and pass it to other subagents as needed.
- Sequence:
  1. Create a basic UI with a text input field and a submit button. (REQUIRED: Create UI task)
  2. Retrieve the user input from the text field when the submit button is clicked. (REQUIRED: Retrieve user input task)
  3. Store the user input as a string variable. (Task 9: Save content to server)
  4. Pass the niche/topic description to other subagents that require it.

F) SUBAGENT 2: Google Image Scraper
- Name: Google Image Scraper
- Final Output: A collection of reference images of top-selling birthday cards in the specified niche, stored as image files or URLs.
- Accomplishment: This subagent will perform a Google Image search using the niche/topic description provided by the Niche Input Handler. It will scrape the search results to obtain reference images of top-selling birthday cards in the niche. The subagent will analyze these images to understand the common visual styles, layouts, and themes used in the niche.
- Sequence:
  1. Receive the niche/topic description from the Niche Input Handler.
  2. Construct a Google Image search query using the niche/topic description (e.g., "{niche topic} birthday cards"). (Task 1: GPT4)
  3. Perform the Google Image search using Task 19: Search Google images.
  4. Extract the URLs of the top reference images from the search results. (Task 1: GPT4)
  5. Analyze the reference images to identify common visual elements, styles, layouts, and themes. (Task 5: Extract visual data from image)
  6. Store the reference image URLs for later use by other subagents. (Task 9: Save content to server)

G) SUBAGENT 3: Niche Researcher
- Name: Niche Researcher
- Final Output: Relevant information, keywords, and phrases related to the specified niche/topic, stored as structured data (e.g., JSON, XML, or a database).
- Accomplishment: This subagent will research the given niche/topic to gather relevant information, keywords, and phrases. It will scrape relevant websites, articles, or utilize pre-existing knowledge bases to collect data that can be used for generating slogans and images.
- Sequence:
  1. Receive the niche/topic description from the Niche Input Handler.
  2. Identify relevant websites, articles, or knowledge bases related to the niche/topic. (Task 15: Search Google)
  3. Scrape the identified sources using Task 16: Scrape text content of website to extract relevant text content.
  4. Perform text analysis and natural language processing (NLP) on the extracted content to identify important keywords, phrases, and relevant information. (Task 1: GPT4)
  5. Store the extracted keywords, phrases, and relevant information as structured data (e.g., JSON, XML). (Task 9: Save content to server)

H) SUBAGENT 4: Slogan Generator
- Name: Slogan Generator
- Final Output: A list of funny slogans or puns based on the niche research, stored as a text file.
- Accomplishment: This subagent will generate a list of funny slogans or puns using the keywords, phrases, and relevant information provided by the Niche Researcher. It will leverage the power of GPT-4 to create humorous and niche-specific slogans for the birthday cards.
- Sequence:
  1. Retrieve the structured data containing keywords, phrases, and relevant information from the Niche Researcher. (Task 10: Access text files already saved)
  2. Use GPT-4 (Task 1) to generate a list of funny slogans or puns based on the retrieved data. The prompt

claude 3 - mermaid-for-all

claude 3 - subagent1

subagent1-refined

A) SUBAGENT SUMMARY: The Niche Input Handler subagent creates a simple user interface that allows the user to enter a description of the niche audience or topic for the birthday cards, stores the user input as a string variable, and passes it to other subagents as needed.

B) FINAL TASK OUTPUT: The niche/topic description entered by the user, stored as a string variable (plain text, up to 1000 characters).

C) SUBAGENT INPUT: None (the subagent starts the workflow by accepting user input through a UI).

E) SUBAGENT TASK SUMMARY:
1. Create a basic UI with a text input field and a submit button. (REQUIRED: Create UI task)
   - Input: None
   - Output: UI with text input field and submit button

2. Retrieve the user input from the text field when the submit button is clicked. (REQUIRED: Retrieve user input task)
   - Input: User interaction with the UI (clicking the submit button)
   - Output: User input as plain text (up to 1000 characters)

3. Use GPT-4 (Task 1) to validate and clean the user input, ensuring it is a coherent and relevant niche/topic description.
   - Input: User input as plain text
   - Output: Validated and cleaned niche/topic description as plain text

4. Store the validated and cleaned user input as a string variable. (Task 9: Save content to server)
   - Input: Validated and cleaned niche/topic description as plain text
   - Output: String variable saved on the server (e.g., server/_public/timestamp/niche_input.txt)

5. Pass the stored niche/topic description URL to other subagents that require it.
   - Input: URL of the stored niche/topic description (e.g., server/_public/timestamp/niche_input.txt)
   - Output: None (the subagent's task is complete)

F) SILOS:
1. UI Creation Silo:
   - Create a basic UI with a text input field and a submit button. (REQUIRED: Create UI task)

2. User Input Retrieval Silo:
   - Retrieve the user input from the text field when the submit button is clicked. (REQUIRED: Retrieve user input task)
   - Use GPT-4 (Task 1) to validate and clean the user input, ensuring it is a coherent and relevant niche/topic description.

3. Data Storage and Sharing Silo:
   - Store the validated and cleaned user input as a string variable. (Task 9: Save content to server)
   - Pass the stored niche/topic description URL to other subagents that require it.

random-autocomplete

To get the end goal of generating a collection of birthday card images with funny, relevant, and niche-specific slogans or puns, along with AI-generated hand-drawn-style images, the most relevant tasks might be:

1. Receive input from the user specifying the niche audience or topic for the birthday cards.
2. Perform a Google image search for "{niche topic} birthday cards" to analyze current top-selling images in the niche and gather inspiration for the AI-generated images and layouts.
3. Research the niche topic using web scraping, API calls, or database queries to gather relevant information, keywords, and phrases.
4. Generate a list of funny, relevant, and niche-specific slogans or puns based on the gathered information using a language model or pun generation algorithm.
5. Use an AI image generation tool (e.g., DALL-E) to create hand-drawn-style images that match the niche topic and complement the generated slogans.
6. Combine the generated slogans and images into birthday card layouts using a template or image manipulation library.
7. Save the final birthday card designs as JPEG images.
8. Return the collection of generated birthday card images to the user.

The overall sequence would be:
Input Niche Topic > Research Niche (Google Images + Web Scraping/APIs) > Generate Slogans > Generate Images > Combine Slogans and Images into Card Layouts > Save as JPEG > Return Results to User

To accomplish this, the Agent could utilize various APIs and libraries, such as:
- Google Custom Search API for performing image searches
- Beautiful Soup or Scrapy for web scraping
- OpenAI GPT-3 or a similar language model for generating slogans and puns
- DALL-E or another AI image generation tool for creating hand-drawn-style images
- Pillow (Python Imaging Library) for combining text and images into card layouts
- os and io libraries for saving the final JPEG images

The Agent, running on a Windows 11 Virtual Machine, would orchestrate the entire process by calling the necessary APIs, libraries, and tools in the appropriate sequence. The Agent's access to a wide range of APIs and Python libraries should make it capable of handling the majority of the tasks required to generate the personalized, niche-specific birthday cards.

subagent1refined-repeating

B) FINAL TASK OUTPUT: The niche/topic description entered by the user, stored as a string variable (plain text, up to 1000 characters).

C) SUBAGENT INPUT: None (the subagent starts the workflow by accepting user input through a UI).

E) SUBAGENT TASK SUMMARY:
1. Create a basic UI with a text input field and a submit button. (REQUIRED: Create UI task)
   - Input: None
   - Output: UI with text input field and submit button

F) SILOS:
1. UI Creation Silo:
   - Create a basic UI with a text input field and a submit button. (REQUIRED: Create UI task)

G) FULL TASKS (INCLUDING REPEATING TASKS):
1. Create a basic UI with a text input field and a submit button. (REQUIRED: Create UI task)
   - Input: None
   - Output: UI with text input field and submit button

In this case, there are no explicit instructions for repeating tasks within the subagent. The tasks are linear and do not require repetition to achieve the desired output.

task-ID-numbers

subagent1-output-variables

Here's the updated task list with the output tokens added:

H) OUTPUT TOKENS

1. Create a basic UI with a text input field and a submit button. (REQUIRED: Create UI task)
   - Input: None
   - Output: [niche-input-ui]

2. Retrieve the user input from the text field when the submit button is clicked. (REQUIRED: Retrieve user input task)
   - Input: [niche-input-ui]
   - Output: [raw-niche-description]

3. Use GPT-4 (Task 1) to validate and clean the user input, ensuring it is a coherent and relevant niche/topic description.
   - Input: [raw-niche-description]
   - Output: [cleaned-niche-description]

4. Store the validated and cleaned user input as a string variable. (Task 9: Save content to server)
   - Input: [cleaned-niche-description]
   - Output: [stored-niche-description-url]

5. Pass the stored niche/topic description URL to other subagents that require it.
   - Input: [stored-niche-description-url]
   - Output: None (the subagent's task is complete)

In this case, the output tokens are:

1. [niche-input-ui]: The basic UI with a text input field and a submit button.
2. [raw-niche-description]: The raw user input from the text field when the submit button is clicked.
3. [cleaned-niche-description]: The validated and cleaned niche/topic description after processing with GPT-4.
4. [stored-niche-description-url]: The URL of the stored niche/topic description string variable on the server.

These output tokens can be used as inputs for other tasks or subagents in the workflow, allowing for a seamless connection between different components of the birthday card generation process.

subagent1refined-repeating-json-core

G) TASKFLOW SUMMARY

1. Title task: Create a basic UI with a text input field and a submit button
   Description of the task: Design and implement a simple user interface that allows users to enter a niche/topic description for birthday cards, with a text input field and a submit button.
   Input(s) involved: None
   Output token name: UI with text input field and submit button

2. Title task: Retrieve the user input from the text field when the submit button is clicked
   Description of the task: Capture the user's input from the text field when they click the submit button, and store it as plain text for further processing.
   Input(s) involved: User interaction with the UI (clicking the submit button)
   Output token name: User input as plain text (up to 1000 characters)

3. Title task: Validate and clean the user input using GPT-4
   Description of the task: Process the user's input using GPT-4 to ensure it is a coherent and relevant niche/topic description, and clean the text if necessary.
   Input(s) involved: User input as plain text
   Output token name: Validated and cleaned niche/topic description as plain text

4. Title task: Store the validated and cleaned user input as a string variable
   Description of the task: Save the validated and cleaned niche/topic description as a string variable on the server, using a timestamp to create a unique file name.
   Input(s) involved: Validated and cleaned niche/topic description as plain text
   Output token name: String variable saved on the server (e.g., server/_public/timestamp/niche_input.txt)

5. Title task: Pass the stored niche/topic description URL to other subagents
   Description of the task: Share the URL of the stored niche/topic description with other subagents that require this information to complete their tasks.
   Input(s) involved: URL of the stored niche/topic description (e.g., server/_public/timestamp/niche_input.txt)
   Output token name: None (the subagent's task is complete)

subagent1mermaid

prompts-required

Based on the list of components for the birthday card generation task, here are the unique prompts that need to be written for the agent workflow:

Task: Niche Research
Output of the task: A list of relevant keywords, phrases, and information related to the given niche/topic.
Description: Write a prompt that instructs the LLM to perform research on the given niche/topic. The prompt should ask the LLM to gather relevant keywords, phrases, and information by analyzing websites, articles, or utilizing pre-existing knowledge bases. The output should be a structured list or JSON object containing the niche-specific data.

Task: Slogan Generation
Output of the task: A list of funny slogans or puns based on the niche research.
Description: Write a prompt that provides the LLM with the niche-specific keywords, phrases, and information obtained from the previous research task. The prompt should instruct the LLM to generate a list of humorous and relevant slogans or puns based on the given data. The output should be a list of slogans in text format.

Task: Image Generation Prompts
Output of the task: A list of image prompts for the AI image generator (e.g., DALL-E) based on the niche and generated slogans.
Description: Write a prompt that takes the niche information and generated slogans as input and instructs the LLM to create a list of image prompts. The image prompts should describe the desired hand-drawn style images that match the niche and slogans. The output should be a list of image prompts in text format, ready to be used with the AI image generator.

Task: Card Layout and Composition Instructions
Output of the task: Instructions for composing the birthday card layout using the generated slogans and images.
Description: Write a prompt that provides the LLM with the generated slogans, images, and any relevant layout information obtained from the Google Image search. The prompt should instruct the LLM to generate step-by-step instructions for composing the birthday card layout. The instructions should include details on how to place the slogans as text and integrate the generated images into the layout. The output should be a clear and concise set of instructions in text format.

These prompts cover the main tasks that require interaction with the LLM in the birthday card generation workflow. The other components, such as the Niche/Topic Input, Google Image Search, and Output Collection, can be handled by the AI agent using APIs, web scraping, or file manipulation techniques without needing specific LLM prompts.