What Shall We Build Next?

Describe

Describe your task

Refine

Refine the plan

SubAgents

Review all agents

Deploy

Deploy your agent

SubAgent #1 - Diagram

Expand Diagram

A) SUBAGENT SUMMARY  
This subagent receives a product name and generates a short, scrolling screenshot video (MP4) of the YouTube search results page for that product.

B) FINAL TASK OUTPUT  
• A single MP4 video URL ([youtube-screenshot-video]) showing a 10-second scrolling capture of “https://www.youtube.com/results?search_query={product_name}”.

C) SUBAGENT INPUT  
• variable1 (string) – The product name or keyword for which we want YouTube search results

E) SUBAGENT TASK SUMMARY  
Below is the detailed workflow showing how the subagent input is transformed step-by-step into the final subagent output (i.e. the scrolling screenshot video of the YouTube search page):

1) Receive the input “variable1” (the product name).  
2) Use SKILL #190 (“Write or rewrite text based on instructions”) to construct the YouTube search URL in the exact format:  
   Prompt example:  
   "Please take the product name: [variable1], and rewrite it as a complete YouTube search URL in the format: https://www.youtube.com/results?search_query={productNameHere}. Return only the URL."  
   → Output: a plain text string of the full YouTube search URL.

3) Pass the newly constructed YouTube search URL to SKILL #175 (“Get Screenshot Video Of Website URL”) to generate a 10-second scrolling browser capture.  
   → Output: An MP4 file URL of the scrolling YouTube results page

4) Subagent final output is the MP4 URL ([youtube-screenshot-video]).

F) SILOS  
Since this subagent’s job is relatively straightforward (construct a URL, then capture a video of that URL), we can treat the entire process as a single silo:  
• Silo: Construct YouTube URL → Capture Screenshot Video.

└──(Input: variable1)  
   └──[#190 → Create YouTube URL from variable1] → [#175 → Screenshot Video of that URL]  
       └──(Output: [youtube-screenshot-video])

SubAgent #2 - Diagram

Expand Diagram

A) SUBAGENT SUMMARY  
This subagent is responsible for finding up to five relevant images for the given product name (variable1).

B) FINAL TASK OUTPUT  
An array (list) of up to 5 URLs of relevant images. (Referred to in the larger workflow as [image-urls])

C) SUBAGENT INPUT  
• The product name (variable1), e.g. "DeepSeek R1"

D) SUBAGENT TASK SUMMARY (Step-by-Step Flow)  
1) Receive the input (product name).  
2) Use Skill #187 (Find Relevant Images with Brave Search) to retrieve up to 5 relevant image URLs for that product name.  
   - INPUT to #187 → The product name (variable1).  
   - OUTPUT from #187 → A list (array) of up to 5 relevant image URLs.  
3) Output that list as the final subagent result ([image-urls]).

E) SILOS  
Since this subagent only needs to perform one main action (search for product images) there is essentially a single silo in this workflow:

• Silo 1:  
   - Step 1: (#187) Find Relevant Images with Brave Search → returns up to 5 image URLs.  
   - Step 2: Pass the list forward as the final subagent output ([image-urls]).

SubAgent #3 - Diagram

Expand Flow

A) SUBAGENT SUMMARY  
"NewsScreenshotAgent" is responsible for finding two relevant news articles about the product (variable1), capturing short scrolling videos of each news page, and then returning a final JSON that includes existing media references (website video, YouTube video, images) along with the newly captured news videos.

B) FINAL TASK OUTPUT  
A single JSON structure (e.g., [final-json-output]) that contains:  
{
  "website_video": "URL_of_website_screenshot_video",
  "youtube_video": "URL_of_youtube_screenshot_video",
  "images": ["URL_of_image1", "URL_of_image2", ...],
  "news_videos": ["URL_of_news_video1", "URL_of_news_video2"]
}

C) SUBAGENT INPUT  
1. The product name (variable1), e.g., "DeepSeek R1"  
2. The previously generated screenshot video of the main website (website_screenshot_video)  
3. The previously generated screenshot video of the YouTube results (youtube_screenshot_video)  
4. The previously gathered images array (image_urls)

E) SUBAGENT TASK SUMMARY  
Below is the step-by-step workflow for the subagent to produce its final JSON output:

1) ASK THE ORACLE FOR NEWS ARTICLES  
   • Skill #224 (Oracle Ask A Question)  
     INPUT:  
       - A prompt such as:  
         "What are the top recent news articles about [variable1]? Please return a summary of the news articles and include the URLs."  
     OUTPUT:  
       - A text response with references/URLs to news articles (oracle-news-summary).

2) PARSE THE ORACLE RESPONSE AND SELECT TWO NEWS URLs  
   • Skill #223 (Powerful LLM Prompt-to-Text Response)  
     INPUT:  
       - The oracle-news-summary (text)  
       - Instruction to extract exactly two relevant article URLs.  
     OUTPUT:  
       - A short text output with two extracted news article URLs (oracle-news-urls).

3) CAPTURE SCROLLING SCREENSHOT VIDEO FOR EACH ARTICLE URL (2x)  
   For each of the two extracted news article URLs:  
   • Skill #175 (Get Screenshot Video Of Website URL)  
     INPUT:  
       - The article URL  
     OUTPUT:  
       - An MP4 video URL showing a short browser scroll capture of the news article.  
     This process runs twice (once for each URL), producing:  
       - news-screenshot-video1  
       - news-screenshot-video2

4) COLLATE ALL REFERENCES INTO A SINGLE FINAL JSON  
   • Skill #223 (Powerful LLM Prompt-to-Text Response)  
     INPUT:  
       - The previously captured references:  
         1) website_screenshot_video  
         2) youtube_screenshot_video  
         3) image_urls  
         4) news-screenshot-video1  
         5) news-screenshot-video2  
       - Instruction/prompt:  
         "I have these media references. Please return them in a JSON structure with keys:  
         {  
           'website_video': ...,  
           'youtube_video': ...,  
           'images': [...],  
           'news_videos': [...]  
         }"  
     OUTPUT:  
       - A final JSON text block containing all URLs (final-json-output).

F) SILOS  
SILO 1: FIND AND EXTRACT NEWS ARTICLE URLs  
1. (Skill #224) → oracle-news-summary  
2. (Skill #223) → oracle-news-urls

SILO 2: CAPTURE NEWS SCROLLING VIDEOS & COMPILE FINAL JSON  
3. (Skill #175) for news URL #1 → news-screenshot-video1  
4. (Skill #175) for news URL #2 → news-screenshot-video2  
5. (Skill #223) to combine all references → final-json-output

That completes the "NewsScreenshotAgent" subagent flow.

4 Template & Links

Expand Flow

A) SUBAGENT SUMMARY  
This subagent’s sole purpose is to collect all media references produced by the previous subagents (website screenshot video, YouTube screenshot video, product images, and two news-article screenshot videos) and compile them into a single JSON output.

B) FINAL TASK OUTPUT  
A single JSON file (for instance, “final_output.json”) containing:  
• "website_screenshot_video" : URL of the main website’s screenshot-video  
• "youtube_screenshot_video" : URL of the YouTube search screenshot-video  
• "image_urls" : array of image URLs  
• "news_screenshot_videos": array of two screenshot-video URLs for top news articles

C) SUBAGENT INPUT  
This subagent requires, as inputs, the final outputs from the other subagents:  
• website_screenshot_video (MP4 URL)  
• youtube_screenshot_video (MP4 URL)  
• image_urls (array of up to 5 image URLs)  
• news_screenshot_video1 (MP4 URL)  
• news_screenshot_video2 (MP4 URL)

E) SUBAGENT TASK SUMMARY (Steps + Skills)  
1) Receive the following inputs:  
   • website_screenshot_video (MP4 URL)  
   • youtube_screenshot_video (MP4 URL)  
   • image_urls (array)  
   • news_screenshot_video1, news_screenshot_video2 (MP4 URLs)

2) Use skill #223 (Powerful LLM Prompt-to-Text Response) to compile these inputs into a single JSON structure.  
   • INPUT to #223:  
     A textual prompt that includes all media references along with an instruction to produce a JSON containing:  
       {
         "website_screenshot_video": <website_screenshot_video>,
         "youtube_screenshot_video": <youtube_screenshot_video>,
         "image_urls": <image_urls array>,
         "news_screenshot_videos": [
           <news_screenshot_video1>,
           <news_screenshot_video2>
         ]
       }
   • OUTPUT of #223: The assembled JSON string.

3) That JSON string becomes the final output of this subagent.

F) SILOS  
This subagent does not require separate silos; it performs one focused task to generate the final JSON.

5 Template & Links

Expand Flow

A) SUBAGENT SUMMARY:  
This subagent’s role is to collect all media references (website screenshot video, YouTube screenshot video, product images, and news article screenshot videos) from the prior subagents and compile them into a single JSON output.

B) FINAL TASK OUTPUT:  
A single JSON object (e.g., saved as “final_output.json”) containing:  
• "website_screenshot_video" (string): MP4 video URL of the main website scrolling capture  
• "youtube_screenshot_video" (string): MP4 video URL of the YouTube search page scroll  
• "images" (array of strings): Up to 5 relevant image URLs  
• "news_screenshot_videos" (array of strings): Exactly 2 MP4 video URLs (for top news articles)

C) SUBAGENT INPUT:  
• website_screenshot_video (string)  
• youtube_screenshot_video (string)  
• images (array of up to 5 URLs)  
• news_screenshot_videos (array of 2 URLs)

E) SUBAGENT TASK SUMMARY (step-by-step workflow):

1) Receive the following four pieces of data from upstream subagents:  
   • website_screenshot_video  
   • youtube_screenshot_video  
   • images (array)  
   • news_screenshot_videos (array)

2) Use SKILL #223 (Powerful LLM Prompt-to-Text Response) to compile these values into one JSON object.  
   • INPUT to skill #223:  
     A prompt instructing the LLM to produce JSON with the exact fields needed.  
     For example:  
     ----------------------------------------------------------------  
     “Please consolidate the following media references into valid JSON with the following structure:  
       {  
         "website_screenshot_video": "...",  
         "youtube_screenshot_video": "...",  
         "images": [...],  
         "news_screenshot_videos": [...]  
       }  
     Here are the values:  
       Website Video: [website_screenshot_video]  
       YouTube Video: [youtube_screenshot_video]  
       Image URLs: [images array]  
       News Videos: [news_screenshot_videos array]

Return only the JSON, with no extra commentary.”  
     ----------------------------------------------------------------

• OUTPUT from skill #223:  
     A JSON string containing the final structure with the references included.

3) SUBAGENT OUTPUT:  
   The JSON string produced by step (2), which will be saved or returned as “final_output.json.”

F) SILOS:  
Because this subagent’s entire purpose is simply to transform previously gathered links/URLs into a single JSON, there is only one silo:  
• Silo 1: Gathering all existing references and converting them via SKILL #223 into final JSON.