What Shall We Build Next?

Describe

Describe your task

Refine

Refine the plan

SubAgents

Review all agents

Deploy

Deploy your agent

A) SUBAGENT SUMMARY  
The “NicheScouter” subagent researches a niche/topic by gathering both textual and visual references (e.g., trending birthday card styles), then condenses all research into a concise (max ~500 characters) summary highlighting jokes, themes, and visual elements relevant to that niche.

B) FINAL TASK OUTPUT  
[niche-research] → A concise text summary (up to ~500 characters) describing popular jokes, themes, or visual styles for the given niche.

C) SUBAGENT INPUT  
{variable1} → The initial description of the niche or topic.

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

1) Use skill #216 (Research Topic (Or Keyword) Deeply)  
   • INPUT: {variable1} (the niche/topic)  
   • OUTPUT: A fairly detailed research text (1,000–3,000 characters) describing the niche.  
   → Let’s call this intermediate result “research-long-text.”

2) Use skill #187 (Find Relevant Images with Brave Search)  
   • INPUT: A keyword query such as "{variable1} birthday cards"  
   • OUTPUT: A list of 5 relevant image URLs that show popular or top-selling birthday cards in the niche.  
   → Let’s call this result “niche-card-images.”

3) Use skill #177 (Analyze 4x Images With GPTVision And Return 4x Text)  
   • Although the name says “4x,” the description allows analyzing 2–18 images; we can pass all 5 images.  
   • INPUT: “niche-card-images” plus a short prompt like “Describe the jokes/themes/visual style from each card image.”  
   • OUTPUT: A text analysis of each image’s style, themes, or unique aspects.  
   → Let’s call this result “image-analysis-text.”

4) Use skill #185 (Write Text (Or Copy) From Inputted Text)  
   • INPUT: Combine “research-long-text” + “image-analysis-text” + instructions to produce a concise ~500-character summary focusing on popular jokes, themes, and visual styles relevant to the niche.  
   • OUTPUT: A final short text summary (≤500 characters).

5) Subagent Final Output → [niche-research]  
   • The concise summary of the niche’s comedic references, popular themes, and relevant visual styles.

F) SILOS

• SILO 1: Textual Research  
  - #216 with {variable1} → “research-long-text”

• SILO 2: Image Research & Analysis  
  - #187 with query “{variable1} birthday cards” → “niche-card-images”  
  - #177 with “niche-card-images” → “image-analysis-text”

• SILO 3: Summarize Everything  
  - #185 with “research-long-text” + “image-analysis-text” → [niche-research] (concise 500-char output)

SubAgent #1 - Diagram

Expand Diagram

A) SUBAGENT SUMMARY:  
"PunSmith" is responsible for producing five short, niche-relevant, pun-based slogans in a single text output.

B) FINAL TASK OUTPUT:  
[pun-slogans] → A single text block containing exactly 5 pun-based slogans.

C) SUBAGENT INPUT:  
• {variable1} (the original user niche/topic description)  
• [niche-research] (short summary from Subagent 1: NicheScouter)

D) SUBAGENT TASK SUMMARY (Detailed Steps):  
1) Action: Use Skill #186 - Create 2–10 Short MicroCopy Text  
   → INPUT:  
      a) The content of [niche-research] (up to ~500 characters)  
      b) The content of {variable1}, to keep the puns relevant to the user’s original input/angle  
      c) Clear instruction to produce exactly 5 short pun-based slogans, referencing both the user’s initial niche description and the research summary

→ OUTPUT: [pun-slogans], which is one text output containing 5 pun-based slogans.

2) Subagent 2’s final output is [pun-slogans].

E) SILOS:  
Since this subagent only needs one skill invocation to generate the final text (the 5 puns), there are no further silos needed.

─────────────────────────
END OF SUBAGENT 2 FLOW

SubAgent #2 - Diagram

Expand Diagram

A) SUBAGENT SUMMARY  
"CardCreatr" takes the pun-based slogans from PunSmith and turns each slogan into a simple hand-drawn-style birthday card image, then returns a final list of those image URLs.

B) FINAL TASK OUTPUT  
[card-image-urls] → A list (3–5) of PNG image URLs, each containing one of the pun slogans in a “hand-drawn cartoon style with bright colors.”

C) SUBAGENT INPUT  
• [pun-slogans]: a text block with 3–5 pun-based slogans (generated by PunSmith).

E) SUBAGENT TASK SUMMARY  
Below is the step-by-step breakdown of how Subagent 3 "CardCreatr" will produce the final output:

1) Receive [pun-slogans] as the input array/list of puns.  
2) For each pun in [pun-slogans]:  
   • Use Skill #222 (Make Image With Text) to generate a single birthday card image in a “hand-drawn cartoon style with bright colors.”  
     - Input to #222:  
       – The pun text (e.g., “Wishing you a woof-tastic day!”)  
       – Style instructions: “hand-drawn cartoon style with bright colors”  
     - Output of #222 for each pun: One PNG image URL with the pun text embedded.  
3) Collect these 3–5 image URLs.  
4) Use Skill #223 (Powerful LLM Prompt-to-Text Response) to combine or neatly list the final figure of URLs into one text block.  
   • Input to #223: The 3–5 PNG image URLs plus an instruction to format them into a concise final list or JSON-style array (which becomes the final subagent output).  
   • Output of #223: A single text block or JSON-like list referencing the generated image URLs.  
5) Return this compiled list (or block) of image URLs as [card-image-urls].

F) SILOS

• Silo 1: Card Generation Steps  
  – Input: [pun-slogans]  
  – Action n1: For each pun, invoke Skill #222 → output a PNG image URL

• Silo 2: Image URL Consolidation  
  – Input: The 3–5 PNG image URLs produced in Silo 1  
  – Action n2: Use Skill #223 to create a final text listing or array of URLs  
  – Output: [card-image-urls]

SubAgent #3 - Diagram

Expand Flow

A) SUBAGENT SUMMARY  
“DesignInspo” – This subagent searches for and analyzes popular niche birthday card designs so we can see existing style trends and gain creative inspiration.

B) FINAL TASK OUTPUT  
[design-inspiration] – A consolidated set of up to 5 reference image URLs (from a Google/Brave-style search) plus a short text summary describing the key design elements or trends observed.

C) SUBAGENT INPUT  
{variable1} – The niche/topic description (e.g., “For dog lovers,” “For Star Wars fans,” etc.)

D) SUBAGENT TASK SUMMARY  
1. Subagent receives {variable1} (the niche/topic).  
2. ACTION STEP → Use #187 (Find Relevant Images with Brave Search)  
   • Input: A query string such as "{variable1} birthday cards"  
   • Output: A list of five relevant external image URLs (search results).

3. ACTION STEP → Use #177 (Analyze 4x Images With GPT Vision & Return 4x Text) (Note: If 5 images are found, you can select the top 4 for a concise analysis, or call #177 in batches–the main idea is to analyze them.)  
   • Input: The 4 (or 5) image URLs and a small prompt like “Describe the style, colors, text usage, and humor elements in these images.”  
   • Output: A short textual commentary about each image’s key design features.

4. ACTION STEP → Use #223 (Powerful LLM Prompt-to-Text Response)  
   • Input: The short commentaries from #177. Instruct the LLM to unify them into a concise summary (1–3 short paragraphs) of observed design patterns/trends.  
   • Output: A single text block summarizing the design traits found in these images.

5. Subagent packages final output as [design-inspiration], containing:  
   • The up to 5 reference image URLs.  
   • The short final summary discussing design elements.

E) SILOS

• SILO 1: IMAGE COLLECTION  
  - Skill #187 searches for “{variable1} birthday cards,” returning up to 5 images.

• SILO 2: IMAGE ANALYSIS  
  - Skill #177 analyzes those returned images to observe and summarize style, layout, color usage, etc.

• SILO 3: FINAL CONSOLIDATION  
  - Skill #223 unifies and condenses the analysis into a concise written summary.  
  - Produces final subagent output [design-inspiration].