What Shall We Build Next?

Describe

Describe your task

Refine

Refine the plan

SubAgents

Review all agents

Deploy

Deploy your agent

A) SUBAGENT SUMMARY  
"Google Search Query Agent" takes a user’s inputs (business type + location), formulates a query, and uses live web sources to retrieve relevant business listings. It returns unstructured text containing details about each business.

B) FINAL TASK OUTPUT  
Unstructured text (raw text) detailing the business listings (business name, address, phone, ZIP, website). This text is not yet structured or cleaned; it is simply the raw information that was discovered.

C) SUBAGENT INPUT  
• businessType (string) – e.g. "plumbers"  
• location (string) – e.g. "Dallas Texas"

E) SUBAGENT TASK SUMMARY  
1) The subagent receives the user’s inputs (businessType, location).  
2) (Optional) It may first assemble a neat query text if needed.  
3) It then calls Skill #224 (Oracle Ask A Question), providing a prompt such as:  
   "Find [BUSINESS_TYPE] in [LOCATION] with name, address, phone, ZIP, website. Return any relevant details in text form."  
4) Skill #224 returns unstructured text about businesses that match the criteria (e.g. a textual list of possible businesses).  
5) The subagent returns this unstructured text (called [raw-business-search-results]) as the subagent’s final output.

F) SILOS

• SILO 1: Query Construction (Optional)
  - (Skill #185 or #190 could be used if needed to create a specific query string.)  
  - Input: (businessType, location) plus instructions to build the query.  
  - Output: "Find [BUSINESS_TYPE] in [LOCATION] with name, address, phone, ZIP, and website."

• SILO 2: Oracle Query & Output
  - (Skill #224: Oracle Ask A Question)  
  - Input: The query text from SILO 1 (or constructed on the fly).  
  - Output: Unstructured text containing relevant business listings.

• SUBAGENT FINAL OUTPUT: [raw-business-search-results] (unstructured text).

SubAgent #1 - Diagram

Expand Diagram

A) SUBAGENT SUMMARY  
This subagent takes raw, unstructured text about various businesses (as returned by Subagent 1) and transforms it into a structured textual list, where each business has clearly defined fields: Business Name, Street Address, City, ZIP, Phone Number, and (optionally) Website.

B) FINAL TASK OUTPUT  
A textual list of businesses, each entry containing:  
• Name  
• Street Address  
• City  
• ZIP  
• Phone  
• Website (if available)

C) SUBAGENT INPUT  
• The unstructured text containing information about multiple businesses, coming from Subagent 1's output.  
• Brief instructions on how to parse that text (e.g., "Parse each business found in the text and return the following fields...").

E) SUBAGENT TASK SUMMARY  
Below is the step-by-step workflow (using the available skills) that the subagent will execute:

1) Subagent Final Input(s):  
   • [raw-business-search-results] (unstructured text listing businesses)  
   • Parsing instructions (e.g. "Extract each business’s details...")

2) SKILL/ACTION #1 (Skill #190 → Write or rewrite text based on instructions):  
   • Input to this skill:  
     - The raw business search results text.  
     - A prompt such as:  
       “Clean up and segment the raw data so each business listing is clearly identified as a separate block or line item. Remove any extraneous text that doesn’t describe a business. Return the cleaned text.”  
   • Output:  
     - [cleaned-block-based-text]: A cleaned version of the unstructured text, with each business listing segregated as its own line or small block.

3) SKILL/ACTION #2 (Skill #190 → Write or rewrite text based on instructions):  
   • Input to this skill:  
     - The [cleaned-block-based-text] from the previous step.  
     - A prompt such as:  
       “For each business block, carefully parse out the following fields if present: (1) Business Name, (2) Street Address, (3) City, (4) ZIP, (5) Phone, (6) Website URL. Then produce a final textual list, where each line includes the six fields in braces, for example:  
         {Name: X, Address: X, City: X, ZIP: X, Phone: X, Website: X}  
       Put 'Website' as blank if the website cannot be found.”  
   • Output:  
     - [extracted-business-details]: A textual list of businesses, each with properly identified Name, Address, City, ZIP, Phone, and Website if present.

4) Subagent Final Output(s):  
   • [extracted-business-details]

F) SILOS

• SILO 1: Cleanup & Segmentation  
  - Uses Skill #190 to remove extraneous text, isolate individual businesses as blocks/lines.  
  - Output: [cleaned-block-based-text]

• SILO 2: Field Extraction  
  - Uses Skill #190 again to parse each block and extract Name, Street Address, City, ZIP, Phone, Website.  
  - Output: [extracted-business-details]

–––––––––––––––––––––––––––––––––––  
This completes the detailed flow for Subagent 2: “Business Data Extraction Agent.”

SubAgent #2 - Diagram

Expand Diagram

A) SUBAGENT SUMMARY:  
This “Data Structuring Agent” receives raw extracted business details (plain text in a loosely structured format) and rewrites them into a consistent, line-by-line dataset—ready to be turned into CSV.

B) FINAL TASK OUTPUT:  
A single block of text (the “structured-business-data”) where each line represents one business record in a consistent field order (City, Business Name, Phone, Street Address, ZIP Code, Website URL), separated by commas.

C) SUBAGENT INPUT:  
The agent’s initial input is the plain-text output from Subagent 2 (“[extracted-business-details]”), containing business name, address, phone, etc., but not yet uniformly organized.

E) SUBAGENT TASK SUMMARY:  
1. The subagent receives “[extracted-business-details]” (unstructured/loosely structured text).  
2. The subagent calls SKILL #190 (Write or rewrite text based on instructions) with a prompt instructing it to parse the input text and produce consistent, comma-separated lines in the order below for each business:  
   • City  
   • Business Name  
   • Phone Number  
   • Street Address  
   • ZIP Code  
   • Website URL  
3. The text response from SKILL #190 will be the final output for Subagent 3, labeled as “[structured-business-data].”

F) SILOS:

• Silo 1: Rewriting into a Structured Format  
  (a) INPUT: “[extracted-business-details]”  
  (b) SKILL #190 → Prompt: “For each business entry within the provided text, extract and order the data into a single line, where fields are: ‘City, Business Name, Phone Number, Street Address, ZIP Code, Website URL.’ Use commas to separate the fields, and place each business on a new line.”  
  (c) OUTPUT: “[structured-business-data]” (the final structured text output).

SubAgent #3 - Diagram

Expand Flow

A) SUBAGENT SUMMARY  
"CSV Generator Agent" takes the cleaned, line-by-line business data (from Subagent 3) and produces a properly formatted CSV text file with a header row.

B) FINAL TASK OUTPUT  
Plain-text CSV (.csv) containing the columns: City, Business Name, Phone Number, Street Address, ZIP Code, Website URL.

C) SUBAGENT INPUT  
The input is the uniform, line-by-line structured business data (e.g., lines of text with fields in a consistent order) that needs to be written out as CSV with the columns and header listed above.

E) SUBAGENT TASK SUMMARY  
1) Subagent receives the structured business data from Subagent 3.  
2) Use skill #190 (Write or rewrite text based on instructions) to:  
   • Read the structured data.  
   • Insert a header row: “City,Business Name,Phone Number,Street Address,ZIP Code,Website URL”.  
   • Convert each line of structured data into a comma-separated row under these headers.  
3) Return the final wording as a single blob of text representing CSV content.  
4) The subagent outputs plain-text CSV content ready to be saved or exported as “CrescentTechnical_Business_Search.csv”.

F) SILOS  
This subagent has a single silo, which is directly transforming the structured data into CSV format via one step:  
• SILO 1: Generate CSV via skill #190 (rewriting text into CSV with the correct header fields).

4 Template & Links

Expand Flow

A) SUBAGENT SUMMARY  
This subagent is responsible for taking the plain-text CSV output from Subagent 4 and presenting it to the user in a simple, final text form so that they can copy/paste or save it as a .csv file (and then open in Excel).

B) FINAL TASK OUTPUT  
The final output is a single text output (referred to as [final-csv-export]) containing:  
• A short user-friendly message and brief instructions on how to use/convert the included CSV text into an Excel file.  
• The actual CSV data (header row and rows of data).

C) SUBAGENT INPUT  
• [structured CSV text] from Subagent 4 (e.g., "City,Business Name,Phone Number,Street Address,ZIP Code,Website URL\nDallas,ABC Plumbing,...").

E) SUBAGENT TASK SUMMARY  
1) Receive the plain-text CSV data as input from Subagent 4.  
2) Use Skill #190 (Write or rewrite text based on instructions) to embed that CSV data into a short, user-friendly text block which includes:  
   • A greeting or short note indicating that this CSV is provided by Crescent Technical.  
   • Instructions that the user can copy/paste the included CSV into a new file or open it in Excel.  
3) Produce the final combined text, labeled as [final-csv-export], which contains both the instructions and the CSV data.

F) SILOS  
• Silo 1: "Combine instructions + CSV data"  
   Inputs: [structured CSV text] from Subagent 4  
   Steps:  
   1) Skill #190 → Rewrite text to include instructions, references to “Crescent Technical,” plus the CSV data.  
   2) Output is the final text that the user can copy/paste or save as “CrescentTechnical_Business_Search.csv.”

No additional silos or sub-steps are needed, since the Export Agent simply returns this final text to the user.

5 Template & Links

Expand Flow

A) SUBAGENT SUMMARY  
This subagent (“Subagent 6”) is responsible for post-processing and validating the structured CSV data before the final spreadsheet is delivered to the user. It ensures that the records (business listings) have valid phone numbers, removes any obvious duplicates, and applies any other final data clean-up steps (e.g., trimming whitespace, ensuring consistent capitalization, etc.).

--------------------------------------------------------------------------------
B) FINAL TASK OUTPUT  
A cleaned, final CSV (or .xlsx) text/string containing the validated records:  
• City  
• Business Name  
• Phone Number  
• Street Address  
• ZIP Code  
• Website URL (if available)

--------------------------------------------------------------------------------
C) SUBAGENT INPUT  
• The structured CSV-like text data (or the text-based CSV representation) that might already have headings and line-by-line entries for each business.

--------------------------------------------------------------------------------
E) SUBAGENT TASK SUMMARY  
Below is the step-by-step detail of how Subagent 6 will process its input and generate its final output. Each arrow “>” indicates output from the previous step being passed as input into the next skill/action.

1. Subagent final input(s)  
   → The CSV text from the previous agent (Subagent 5 or wherever the data is coming from).

2. Validate & Clean Data (#190 - “Write or rewrite text based on instructions”)  
   INPUT NEEDED:  
      • The raw CSV text listing all businesses.  
      • Instructions to:  
         • Split the text by lines.  
         • Discard lines missing a valid business name or phone number (if phone is mandatory).  
         • Remove any obvious duplicates (e.g., two lines with exactly the same Name + Phone + Address).  
         • Normalize phone formats (e.g., strip out spaces, ensure standard parentheses or dashes, etc.).  
         • Trim leading/trailing whitespace from each cell.  
         • Convert city names to title case (e.g., “dallas” → “Dallas”) if desired.  
   OUTPUT:  
      • Cleaned CSV text with valid, deduplicated, and normalized rows.

3. Return Final Clean CSV (#190 - “Write or rewrite text based on instructions”)  
   • Although #190 was used to clean the data, a second (or the same single) prompt might finalize the CSV format, ensuring all rows have the correct number of columns in the correct order:  
       1) City  
       2) Business Name  
       3) Phone Number  
       4) Street Address  
       5) ZIP Code  
       6) Website URL  
   • Output as plain-text CSV so it can be easily downloaded or opened in Excel.

4. Subagent final output(s)  
   • The final validated CSV text / data string (ready to be saved as an .xlsx if desired).

--------------------------------------------------------------------------------
F) SILOS  
Depending on how large or complex the validation is, you might split Subagent 6 into smaller silos:  
1) “Deduplication & Basic Validation Silo” – uses #190 once to identify duplicates, remove them, and remove incomplete rows.  
2) “Format Normalization Silo” – uses #190 (or the same #190 call with specialized instructions) to ensure consistent phone formats, consistent capitalization, removed whitespace, etc.

In practice, these can be done within one or two calls to skill #190 (and possibly repeated if needed), but separating them helps you stay organized if the final CSV requires more complex cleaning rules.