What Shall We Build Next?

Describe

Describe your task

Refine

Refine the plan

SubAgents

Review all agents

Deploy

Deploy your agent

A) SUBAGENT SUMMARY  
DataExtractionBot extracts product details (title, bullet points, description, and main image URL) from an Amazon or Etsy product page.

B) FINAL TASK OUTPUT  
[extracted-product-details] → A single JSON object of the form:  
{  
  "title": "...",  
  "bullets": ["...", "..."],  
  "description": "...",  
  "imageURL": "..."  
}

C) SUBAGENT INPUT  
{product-url}  (the URL to an Amazon or Etsy product page)

D) SUBAGENT OUTPUT  
[extracted-product-details] (a JSON object with the fields listed above)

E) SUBAGENT TASK SUMMARY  
1. Use Skill #226 (Extract Structured Data From 1x URL)  
   • INPUT: {product-url} + an instruction requesting:  
     “Please extract the product’s title, bullet points, description, and main image URL. Return a text summary of these fields.”  
   • OUTPUT: A textual summary that includes the product’s title, bullet points, description, and main image URL.

2. Use Skill #223 (Powerful LLM Prompt-to-Text Response)  
   • INPUT: The textual summary from Step 1, plus a prompt to transform it into strict JSON format with the structure:  
     { "title": "...", "bullets": [...], "description": "...", "imageURL": "..." }  
   • OUTPUT: [extracted-product-details] (the final JSON object).

F) SILOS  
Single silo (just two sequential steps as outlined above).

SubAgent #1 - Diagram

Expand Diagram

A) SUBAGENT SUMMARY  
This subagent (“ComplianceCheckerBot”) examines the extracted text (title, bullets, description, etc.) for any trademark or policy violations, checks suspicious terms against a trademark database, and rewrites/removes flagged terms to produce a final, policy-compliant text output and relevant non-infringing keywords.

B) FINAL TASK OUTPUT  
[compliant-text-and-keywords] → A single structured text block (e.g., JSON) containing the sanitized title, bullet points, product description, and final keywords that are safe to use.

C) SUBAGENT INPUT  
[extracted-product-details] → The text content extracted from the previous subagent (title, bullet points, description, possibly main image info if available).

E) SUBAGENT TASK SUMMARY  
Below is the specific sequence of skill actions to achieve the subagent’s goal:

1) (Input) [extracted-product-details]  
   └─ This contains the full text from the original listing (title, bullets, description).

2) #223 - Powerful LLM Prompt-to-Text Response  
   └─ Task: “Identify any trademark or policy violations found in the listing text. Return all suspicious terms and whether they violate Amazon Merch rules (hateful content, profanity, etc.) or are possibly trademarked.”  
   └─ Output: A text listing of suspicious terms.

3) #224 - Oracle Ask A Question  
   └─ Task: For each suspicious term (if any) from Step 2, check whether it is trademarked (e.g., “{Term} trademark?”).  
   └─ Output: A text summary that identifies which terms are confirmed or likely trademarked.

4) #190 - Write or rewrite text based on instructions  
   └─ Task: Rewrite or remove any flagged or trademarked terms in the original text. Ensure compliance with Amazon’s merch policies. Retain meaningful, non-infringing keywords. Output final sanitized text and final keyword list.  
   └─ Output: [compliant-text-and-keywords] → A structured JSON/text block containing the safe title, bullet points, product description, and safe keywords.

F) SILOS

• SILO 1 (Analysis):  
   - Input the listing text into #223 to identify potential violations or trademark issues.

• SILO 2 (Trademark Check):  
   - For each suspicious term flagged, use #224 to query the trademark database and confirm any matches.

• SILO 3 (Rewrite & Final Output):  
   - Use #190 to rewrite or remove flagged terms.  
   - Produce one consolidated [compliant-text-and-keywords] as the subagent’s final output.

SubAgent #2 - Diagram

Expand Diagram

A) SUBAGENT SUMMARY  
DesignGeneratorBot is responsible for taking the final, cleaned keywords and concepts (from the ComplianceCheckerBot) and producing a brand-new T-shirt design image as a high-resolution PNG (transparent background, 300 DPI, 4500 × 5400 pixels).

B) FINAL TASK OUTPUT  
[new-tshirt-design-png] → A single PNG file:  
• Dimensions: 4500 × 5400 pixels  
• Resolution: 300 DPI  
• Transparent background

C) SUBAGENT INPUT  
[compliant-text-and-keywords] → This is the cleaned, policy-compliant text/keywords output from Subagent 2.

E) SUBAGENT TASK SUMMARY  
Below is the step-by-step workflow for Subagent 3 (DesignGeneratorBot), illustrating which skills/actions are used and in what sequence. The final result is the new high-resolution T-shirt design PNG.

1) #222 - Make Image  
   • INPUT: [compliant-text-and-keywords] (used to form the text prompt instructing the AI to create a T-shirt design in a style reflecting the safe keywords).  
   • TASK: Generate a brand-new T-shirt design image emphasizing that it should feature a transparent background or at least a background that can be removed.  
   • OUTPUT: PNG image URL (initial design).

2) BACKGROUND REMOVAL (External API Call)  
   • INPUT: The PNG output from step 1.  
   • TASK: Use the ai-background-remover API (as provided via custom snippet) to remove any non-transparent area and ensure a transparent background.  
   • OUTPUT: PNG image URL with background removed.

3) AI IMAGE UPSCALER (External API Call)  
   • INPUT: The background-removed PNG from step 2.  
   • TASK: Use the ai-image-upscaler API (as provided via custom snippet) to upscale and help maintain clarity toward a higher resolution.  
   • OUTPUT: Upscaled PNG image URL (still needs final exact dimension).

4) #191 - Resize Image  
   • INPUT: The upscaled PNG from step 3, plus the required final dimensions (4500 × 5400).  
   • TASK: Ensure the new image size is exactly 4500 × 5400 pixels (keeping a transparent background).  
   • OUTPUT: [new-tshirt-design-png] — the final PNG file (4500 × 5400, 300 DPI, transparent).

F) SILOS  
Because the process is a straightforward linear sequence, everything is grouped into one silo:  
• Silo 1: Generate → Remove Background → Upscale → Resize.

This completes Subagent 3’s workflow. The final deliverable is a single high-resolution, transparent PNG design.

SubAgent #3 - Diagram

Expand Flow

A) SUBAGENT SUMMARY  
This subagent (ListingTextCreatorBot) takes the policy-compliant keywords/text from the previous subagent and writes new Amazon listing content (title, bullet points, description) in a single text block.

B) FINAL TASK OUTPUT  
[revised-listing-text] → A single textual output (title, bullet points, description) that is policy-compliant and uses the safe keywords.

C) SUBAGENT INPUT  
[compliant-text-and-keywords] → Clean, policy-compliant text and keywords from Subagent 2.

E) SUBAGENT TASK SUMMARY  
1) Receive [compliant-text-and-keywords] from Subagent 2.  
2) Use Skill #190 - Write or rewrite text based on instructions, prompting it to create:  
   • Title  
   • Bullet points (amazon style)  
   • Description  
   …all using the clean/safe keywords.  
3) Output [revised-listing-text] as a single text block containing the new title, bullet points, and description.

F) SILOS  
No additional silos or subdivisions are needed. This sequence is a single-step process:
[compliant-text-and-keywords] → (#190) → [revised-listing-text]

4 Template & Links

Expand Flow

A) SUBAGENT SUMMARY  
“Subagent 5 (FinalOutputBot) obtains the newly generated T-shirt image and revised listing text, optionally performs a final trademark/policy check, and packages both into a single final deliverable for the user.”

B) FINAL TASK OUTPUT  
[final-package], containing:  
• The final PNG design file (transparent background, 300 DPI, 4500×5400).  
• The final listing text (title, bullet points, description).

C) SUBAGENT INPUT  
[new-tshirt-design-png], [revised-listing-text]

D) SUBAGENT TASK SUMMARY  
1) (Optional) Use SKILL #224 (“Oracle Ask A Question”) – The subagent performs a quick final trademark/policy confirmation on [revised-listing-text].  
   – Input:  
     “Check if the following T-shirt listing text has any obvious trademark or policy concerns for Amazon Merch: [revised-listing-text]. Summarize any issues.”  
   – Output: A short, up-to-date answer describing any potential red flags.

2) Use SKILL #223 (“Powerful LLM Prompt-to-Text Response”) – The subagent compiles the final output for the user, referencing both the [new-tshirt-design-png] and [revised-listing-text].  
   – Input Prompt Example:  
     “Combine the final T-shirt image URL: [new-tshirt-design-png], the final listing text: [revised-listing-text], and if applicable, the summary from the trademark check. Present it as a single final package for the user.”  
   – Output: A final text message or structured text (the [final-package]) that includes the PNG URL and the listing text blocks (title, bullet points, description), plus any remarks from the trademark check.

E) SILOS  
No additional silos. The subagent performs two main actions in sequence (final check → final package), then outputs the final deliverable.