What Shall We Build Next?

Describe

Describe your task

Refine

Refine the plan

SubAgents

Review all agents

Deploy

Deploy your agent

A) SUBAGENT SUMMARY  
"Listing Data Extractor" retrieves the original title, bullet points, description, and main image URL from a single provided listing URL (Amazon or Etsy/other).

B) FINAL TASK OUTPUT  
[extracted-listing-data] → A structured data object containing:  
• title  
• bullet points  
• description  
• main image URL

C) SUBAGENT INPUT  
{listing-url}

E) SUBAGENT TASK SUMMARY  
1. Determine the domain (Amazon vs. other) from {listing-url}.  
2. If Amazon:  
   → Use Skill #225 (Crawl 1x Amazon URL) to extract structured data from the product page.  
   → Output is text summarizing the page data.  
3. If not Amazon:  
   → Use Skill #226 (Extract Structured Data From 1x URL) to parse structured data from the page.  
   → Output is text summarizing the page data.  
4. Convert the raw text output from #225 or #226 into a standardized JSON-like structure.  
   → Use Skill #223 (Powerful LLM Prompt-to-Text Response) with a prompt to parse out “title,” “bullet points,” “description,” and “main image URL.”  
   → Return [extracted-listing-data] as the final single structured object.

F) SILOS

• SILO 1: Domain Check & Extraction  
   1) Skill #223 (Powerful LLM Prompt-to-Text Response):  
      – Input: {listing-url}  
      – Output: Analyzed domain (“amazon” | “other”).  
   2) If domain = “amazon”:  
      – Skill #225 (Crawl 1x Amazon URL)  
         → Input: {listing-url + instructions to get product data}  
         → Output: Summarized text of product data (title, bullet points, description, image).  
      Else (domain ≠ “amazon”):  
      – Skill #226 (Extract Structured Data From 1x URL)  
         → Input: {listing-url + instructions to get product data}  
         → Output: Summarized text of product data (title, bullet points, description, image).

• SILO 2: Structuring Final Data  
   3) Skill #223 (Powerful LLM Prompt-to-Text Response)  
      – Input: “Here is the text from the previous extraction step; please return valid JSON with fields {title, bullet_points, description, main_image_url}.”  
      – Output: Final JSON object → [extracted-listing-data].

SubAgent #1 - Diagram

Expand Diagram

A) SUBAGENT SUMMARY:  
This subagent (Keyword & Image Analysis) takes the previously extracted listing data (title, bullet points, description, main image URL), produces a short descriptive summary of the original shirt’s design (colors, style, patterns), and also identifies/removes trademarked or potentially infringing terms from the text, returning a final structured object with both the “safe” keywords and image description.

B) FINAL TASK OUTPUT:  
[safe-keywords-and-image-description] → A single structured data object with two parts:  
• "ImageDescription": A textual description of the shirt design (colors, style, patterns).  
• "SafeKeywords": A list of keywords (and possibly short phrases) from the original listing text with potentially infringing or trademarked terms removed.

C) SUBAGENT INPUT:  
[extracted-listing-data] → This should contain:  
• "title" (string)  
• "bulletPoints" (array or list of strings)  
• "description" (string)  
• "mainImageUrl" (string)

E) SUBAGENT TASK SUMMARY (Step-by-Step Flow):

1) Analyze the main shirt image.  
   → Skill #176 (Analyze An Image With GPT Vision & Return Text)  
   INPUT: mainImageUrl from [extracted-listing-data], along with a suitable prompt (e.g., “Describe the colors, style, and any visible design elements in this shirt image.”)  
   OUTPUT: (Text) "ImageDescription"

2) Identify and remove trademarked/infringing terms from listing text.  
   → Skill #223 (Powerful LLM Prompt-to-Text Response)  
   INPUT: The full text from the listing (e.g., “title,” “bulletPoints,” “description”) plus instructions to detect and/or remove trademarked or infringing words/phrases.  
   OUTPUT: (Text) "SafeKeywords" (list of non-trademarked, safe words/phrases)

3) Combine “ImageDescription” and “SafeKeywords” into one final structured output.  
   → Skill #223 (Powerful LLM Prompt-to-Text Response)  
   INPUT: The two pieces together: “ImageDescription” from Step 1 and “SafeKeywords” from Step 2, along with instructions to unify them into a single JSON or structured object.  
   OUTPUT: [safe-keywords-and-image-description] (Subagent2Output)

F) SILOS (If Needed):  
• Silo 1: Image Analysis (Step 1)  
• Silo 2: Listing Text Check (Step 2)  
• Silo 3: Final Merge (Step 3)

That completes the Subagent 2 (Keyword & Image Analysis) flow.

SubAgent #2 - Diagram

Expand Diagram

A) SUBAGENT SUMMARY  
“Design Generator & Processor” takes safe keywords and a style description of the original design, then outputs two brand-new t-shirt designs as transparent PNGs at 300 dpi and 4500×5400 pixels.

B) FINAL TASK OUTPUT  
• Two final PNG files, each with transparent background, 300 dpi, and sized 4500×5400.

C) SUBAGENT INPUT  
• [safe-keywords-and-image-description] – text object containing:  
  1) A set of safe, non-trademarked keywords.  
  2) A descriptive summary of the original shirt design (colors, style, patterns).

E) SUBAGENT TASK SUMMARY  
1) Receive input [safe-keywords-and-image-description].  
2) Skill #222 (Make Image): Generate the first design variation using the safe keywords and style description, explicitly requesting a transparent background.  
   → Output: Variation1_raw.png  
3) Skill #191 (Resize Image): Resize Variation1_raw.png to 4500×5400 at 300 dpi.  
   → Output: Variation1_final.png  
4) Skill #222 (Make Image): Generate a second design variation with the same safe keywords and style description, also requesting a transparent background.  
   → Output: Variation2_raw.png  
5) Skill #191 (Resize Image): Resize Variation2_raw.png to 4500×5400 at 300 dpi.  
   → Output: Variation2_final.png

After step 5, the subagent has created two final design variations.

F) SILOS  
• SILO 1: Generate First Variation  
   – #222 → #191 → Variation1_final.png  
• SILO 2: Generate Second Variation  
   – #222 → #191 → Variation2_final.png

The subagent’s final output is [generated-tshirt-image], which in this case consists of two distinct PNGs (Variation1_final.png and Variation2_final.png) returned together.

SubAgent #3 - Diagram

Expand Flow

A) SUBAGENT SUMMARY:  
"Listing Text Creator" takes the safe, non-infringing keywords and the design description from previous subagents, then crafts a concise Amazon-compliant listing (title, bullet points, description) while ensuring no trademarked terms remain.

B) FINAL TASK OUTPUT:  
[final-listing-text] → A text file containing the finalized listing copy:  
• Title (≤60 characters)  
• 2 bullet points (each ≤200 characters)  
• Description (≤1000 characters)

C) SUBAGENT INPUT:  
[safe-keywords-and-image-description]  
(This structured data includes the non-trademarked keywords and a descriptive summary of the original shirt design.)

E) SUBAGENT TASK SUMMARY:  
1. Input: [safe-keywords-and-image-description]  
2. Skill #190 – “Write or rewrite text based on instructions”  
   – Purpose: Generate a draft listing (title, bullet points, description) using the safe keywords.  
   – Output: [draft-listing-text] (must observe Amazon character limits and avoid trademarked terms.)  
3. Skill #223 – “Powerful LLM Prompt-to-Text Response”  
   – Purpose: Check and revise the [draft-listing-text] if any residual trademark terms or policy violations exist, then finalize the listing text.  
   – Output: [final-listing-text]

F) SILOS (if any):  
No separate silos are required. This subagent uses two sequential skills in one flow.