What Shall We Build Next?

Describe

Describe your task

Refine

Refine the plan

SubAgents

Review all agents

Deploy

Deploy your agent

Let me break down the Amazon Product Analyzer subagent in detail:

A) SUBAGENT SUMMARY: 
An automated system that crawls Amazon bestseller pages, extracts product data, and identifies items with the highest potential affiliate commission rates.

B) FINAL TASK OUTPUT:
A structured JSON file containing:
- Product title
- ASIN
- Current price
- Category
- Commission rate
- Calculated potential earnings
- Product URL
- Best seller rank
- Rating and review count

C) SUBAGENT INPUT:
- Amazon product category name (e.g., "Electronics", "Kitchen & Dining")
- Target marketplace (e.g., amazon.com, amazon.co.uk)

D) SUBAGENT TASK SUMMARY:
Input > Skill #218 (Brainstorm category keywords) > Skill #225 (Crawl Amazon bestseller URL) > Skill #216 (Research commission rates) > Skill #223 (LLM analysis for earnings calculation) > Skill #185 (Format data into structured JSON) > Output

E) SILOS:
The subagent operates in 3 distinct silos:

SILO 1: CATEGORY RESEARCH & PAGE IDENTIFICATION
1. Input: Category name
2. Skill #218: Brainstorm Related Keywords
   - Input: Category name
   - Output: List of specific category keywords
3. Skill #224: Oracle Ask A Question
   - Input: Category keywords
   - Output: Verified Amazon bestseller page URL

SILO 2: PRODUCT DATA EXTRACTION
1. Skill #225: Crawl Amazon URL
   - Input: Bestseller page URL
   - Output: Raw product data
2. Skill #216: Research Topic Deeply
   - Input: Product ASINs
   - Output: Commission rate data
3. Skill #223: Powerful LLM Analysis
   - Input: Product data + commission rates
   - Output: Calculated potential earnings

SILO 3: DATA STRUCTURING & OUTPUT
1. Skill #185: Write Text From Input
   - Input: All collected data
   - Output: Formatted JSON structure
2. Skill #227: Return Graph Image
   - Input: Product earnings comparison data
   - Output: Visual earnings comparison chart
3. Final Output: JSON file + visualization

This breakdown ensures systematic data collection, analysis, and structured output while maintaining clear separation of concerns between different functional areas of the subagent.

The JSON output structure is optimized for consumption by subsequent subagents in the workflow, particularly the Review Research Aggregator which will use this data as its input.

SubAgent #1 - Diagram

Expand Diagram

Let me break down this Review Research Aggregator subagent in detail.

A) SUBAGENT SUMMARY: 
A research collection and analysis system that gathers product reviews from multiple sources, processes them for key information, and outputs structured review data in a standardized JSON format.

B) FINAL TASK OUTPUT:
A structured JSON file containing:
- Aggregated review data from multiple sources
- Sentiment analysis scores
- Key product features/benefits
- Common complaints/issues
- Technical specifications
- Comparative analysis points
- Source URLs and citations

C) SUBAGENT INPUT:
- Product name/identifier
- Amazon product URL
- Product category
- Target review count (optional)

E) SUBAGENT TASK SUMMARY:
The workflow chains together as follows:

1. Initial Research Collection:
#216 (Research Topic Deeply) > Output: Comprehensive research text about product

2. Amazon-Specific Data:
#225 (Crawl Amazon URL) > Output: Structured Amazon review data

3. Extract Additional Data Points:
#224 (Oracle Ask Questions) with specific queries:
- "What are the top 3 YouTube reviews for this product?"
- "What are professional reviewers saying about this product?"
- "What are the main technical specifications?"
- "What are common complaints about this product?"
> Output: Detailed answers to specific product questions

4. Process & Structure Data:
#223 (Powerful LLM Prompt-to-Text Response) to analyze and structure all collected data
> Output: Organized text summary

5. Final Data Organization:
#185 (Write Text From Inputted Text) to format all data into proper JSON structure
> Output: Final JSON file with structured review data

F) SILOS:
The subagent operates in three distinct silos:

SILO 1: RAW DATA COLLECTION
- Purpose: Gather initial unstructured review data
- Skills: #216, #225
- Output: Raw review text and structured Amazon data

SILO 2: DETAILED ANALYSIS
- Purpose: Extract specific data points and analysis
- Skills: #224, #223
- Output: Analyzed and categorized review information

SILO 3: DATA STRUCTURING
- Purpose: Format and structure all data into final format
- Skills: #185
- Output: Final JSON file with complete review dataset

This workflow ensures thorough data collection, proper analysis, and structured output that can be easily used by subsequent subagents in the main workflow.

SubAgent #2 - Diagram

Expand Diagram

Let me break down the Media Collector subagent in detail:

A) SUBAGENT SUMMARY: 
An image processing system that collects, optimizes, and prepares product-related images with metadata for use in an affiliate product review article.

B) FINAL TASK OUTPUT:
A directory containing:
- 3-5 high-quality product images in PNG format (1024x1024 for product shots)
- Each image with associated metadata JSON file containing:
  * Image dimensions
  * File size
  * Alt text description
  * Source URL
  * Image content analysis
- All images web-optimized and properly sized for article use

C) SUBAGENT INPUT:
- Product name/description
- Amazon product URL
- Product category/niche

E) SUBAGENT TASK SUMMARY:
1. Image Collection Phase:
   - Use #187 (Brave Search) to find relevant product images
   - Use #225 (Crawl Amazon URL) to extract official product images
   
2. Image Analysis Phase:
   - Use #176 (GPT Vision Analysis) for each image to generate descriptions
   - Use #177 (Batch Image Analysis) to compare and select best images

3. Image Processing Phase:
   - Use #191 (Resize Image) to standardize dimensions
   - Use #221 (Recreate Image) for any low-quality images that need enhancement

4. Metadata Generation Phase:
   - Use #223 (LLM Response) to generate SEO-optimized alt tags
   - Use #185 (Write Text) to create image descriptions

F) SILOS:
SILO 1: IMAGE COLLECTION
Input: Product details
- Skill #225: Crawl Amazon URL for official images
- Skill #187: Search for additional product images
Output: Collection of 8-10 raw image URLs

SILO 2: IMAGE ANALYSIS & SELECTION
Input: Raw image URLs
- Skill #177: Analyze all images in batch
- Skill #223: Generate selection criteria
Output: Top 3-5 selected images with analysis

SILO 3: IMAGE PROCESSING
Input: Selected images
- Skill #191: Resize each image to standard dimensions
- Skill #221: Enhance any lower quality images
Output: Optimized image files

SILO 4: METADATA CREATION
Input: Processed images
- Skill #176: Analyze each final image individually
- Skill #185: Generate detailed metadata text
Output: JSON metadata files for each image

This workflow ensures we collect, analyze, process, and prepare high-quality images with complete metadata, ready for integration into the final article HTML.

SubAgent #3 - Diagram

Expand Flow

I'll analyze SUBAGENT 4 (Content Generator) and break it down according to your guidelines:

A) SUBAGENT SUMMARY:
A specialized content creation system that transforms structured product and review data into a comprehensive, well-formatted markdown review article ready for HTML conversion.

B) FINAL TASK OUTPUT:
A markdown (.md) file containing:
- Complete product review (2000-3000 words)
- Properly formatted headers (H1, H2, H3)
- Product specifications in table format
- Pros/cons lists
- Star ratings (⭐) formatting
- Image placeholders with alt text
- Affiliate link placeholders
- Schema markup instructions

C) SUBAGENT INPUT:
- JSON file containing product details (name, price, specs, etc.)
- JSON file containing aggregated review data from multiple sources
- Array of image URLs with metadata
- Affiliate link data

E) SUBAGENT TASK SUMMARY:
Input > #216 Research Topic Deeply (for additional context) > #223 Powerful LLM for article outline > #185 Write Text From Input (for main review sections) > #190 Write/Rewrite Text (for product specifications) > #186 Create Micro-Copy (for headlines/subheaders) > #190 Write/Rewrite Text (for final markdown formatting) > Output markdown file

F) SILOS:
The subagent operates in 4 distinct silos:

SILO 1: RESEARCH & ANALYSIS
1. #216 Research Topic Deeply
   Input: Product name + category
   Output: Comprehensive research summary
2. #224 Oracle Ask Question (for current product details)
   Input: Specific product questions
   Output: Updated product information

SILO 2: OUTLINE GENERATION
1. #223 Powerful LLM
   Input: Research + review data
   Output: Detailed article outline

SILO 3: CONTENT CREATION
1. #185 Write Text From Input
   Input: Research + outline for main review
   Output: Main review content
2. #190 Write/Rewrite Text
   Input: Product specifications
   Output: Formatted spec tables
3. #186 Create Micro-Copy
   Input: Product highlights
   Output: Headlines and subheaders

SILO 4: MARKDOWN FORMATTING
1. #190 Write/Rewrite Text
   Input: All previous content
   Output: Final formatted markdown file with:
   - Proper markdown syntax
   - Image placeholders
   - Schema markup instructions
   - Affiliate link formatting
   - Table formatting
   - Header hierarchy

This structured approach ensures each component is handled separately but builds toward the final markdown output that's ready for HTML conversion.

4 Template & Links

Expand Flow

Let me break down the HTML Builder subagent in detail:

A) SUBAGENT SUMMARY: 
A specialized HTML assembly agent that takes structured content and media assets and converts them into a fully-formatted, SEO-optimized, responsive HTML product review page.

B) FINAL TASK OUTPUT:
A single index.html file containing:
- Complete HTML5 structure with proper DOCTYPE and meta tags
- Embedded responsive images with alt tags and schema markup
- Structured product review content with proper heading hierarchy
- Schema.org Product Review markup
- Internal CSS styling for responsive design
- Amazon affiliate links with tracking
- SEO meta tags and Open Graph data

C) SUBAGENT INPUT:
- JSON file containing structured product review content
- Directory of processed product images with metadata
- Product pricing and affiliate link data
- SEO keywords and meta information

E) SUBAGENT TASK SUMMARY:
1. Process Input Content (#185)
- Input: JSON review content
- Action: Transform structured data into HTML-ready content blocks
- Output: Formatted HTML content sections

2. Process Images (#191 x 3)
- Input: Product image directory
- Action: Resize images to responsive breakpoints (desktop/tablet/mobile)
- Output: Optimized image set with multiple sizes

3. Generate Schema Markup (#223)
- Input: Product and review data
- Action: Create JSON-LD schema markup for product review
- Output: Schema markup code block

4. Generate SEO Elements (#223)
- Input: Product data and keywords
- Action: Create meta tags and Open Graph data
- Output: HTML header elements

5. Final Assembly (#223)
- Input: All processed components
- Action: Combine all elements into final HTML template
- Output: Complete HTML file

F) SILOS:
SILO 1: CONTENT PREPARATION
- Skill #185: Process review content into HTML blocks
- Skill #223: Generate meta descriptions and SEO elements

SILO 2: MEDIA PROCESSING
- Skill #191 (x3): Create responsive image sets
  - Desktop version (1200px width)
  - Tablet version (768px width)
  - Mobile version (375px width)

SILO 3: TECHNICAL ELEMENTS
- Skill #223: Generate schema markup
- Skill #223: Create meta tags and Open Graph data

SILO 4: FINAL ASSEMBLY
- Skill #223: Combine all elements into final HTML template with:
  - Header section with meta tags
  - Body content with responsive images
  - Schema markup in script tags
  - Internal CSS styling
  - Affiliate link integration

This breakdown ensures each component is properly processed and assembled into a cohesive, SEO-optimized HTML document that meets modern web standards and accessibility requirements.