Web Search Results Agent - Confirm & Copy Final Agent Code

Describe

Describe your task

Refine

Refine the plan

SubAgents

Review all agents

Deploy

Deploy your agent

Review Final Agent Copy

Expand

[
  {
    "taskID": 224,
    "semanticTitleOfTask": "Oracle Web Search to retrieve raw results from the internet",
    "taskDescription": "This task leverages Skill #224 to perform a real-time web search using the user's query string, returning relevant pages, URLs, and snippet data in an unstructured text format.",
    "inputDescription": "This task requires VARIABLE1 (the user’s search query in string format). The query will be passed to the Oracle Ask A Question skill to perform a real-time web search.",
    "inputRequired": [
      "VARIABLE1"
    ],
    "outputDescription": "The output is a text-based response containing raw search results. This includes webpage titles, URLs, and snippets, all returned in an unstructured format.",
    "outputName": "raw_search_results",
    "promptInstruction": ""
  },
  {
    "taskID": 223,
    "semanticTitleOfTask": "Parse raw textual search data into structured JSON array",
    "taskDescription": "This skill-based task uses a language model to parse the raw textual search results and extract each search result's title, URL, and snippet into a structured JSON array. The final output supports further data handling or storage.",
    "inputDescription": "Requires the unstructured text from 'raw_search_results'. It identifies separate entries in the text, then extracts the relevant fields for each search result.",
    "inputRequired": [
      "raw_search_results"
    ],
    "outputDescription": "A JSON-like array of objects, each containing 'title', 'url', and 'snippet'. This output can be readily used or stored by subsequent workflows.",
    "outputName": "parsed_search_results",
    "promptInstruction": "You are a precise data parser. I will provide raw search results in text format. Your task: Identify each search result entry, extracting 'title', 'url', and 'snippet' (if available). Return a valid JSON array with objects containing these fields. If a snippet isn't available, use an empty string for that field. Include no extra commentary or formatting."
  },
  {
    "taskID": 1,
    "semanticTitleOfTask": "Assemble final search results JSON from query and parsed results array",
    "taskDescription": "This task uses skill #223 to merge the user’s search query and an array of parsed search results. It outputs a well-formed JSON object containing the query string along with a 'results' array of objects, each having 'title', 'url', and an optional 'snippet'.",
    "inputDescription": "Requires two inputs: VARIABLE1 (the user's original search query) and parsed_search_results (each object contains 'title', 'url', and optionally 'snippet'). These inputs are used to form the final JSON structure.",
    "inputRequired": [
      "VARIABLE1",
      "parsed_search_results"
    ],
    "outputDescription": "A valid JSON string containing two top-level keys: 'query' (the original user query) and 'results' (an array of objects). Each object includes a 'title', a 'url', and potentially a 'snippet' field. The JSON is suitable for saving as 'search_results.json'.",
    "outputName": "assembled_json_file",
    "promptInstruction": "You are a JSON formatting specialist. Using the provided search query and results array, create a single JSON object with two top-level keys: 'query' and 'results'. The 'query' value should be the exact search query string provided. The 'results' value should be an array containing the provided search result objects. Each result object must have a 'title' and 'url' field, and may optionally include a 'snippet' field. Format the JSON with proper indentation. Return only the JSON string with no additional commentary or explanation. Ensure the output is valid JSON that could be parsed by any standard JSON parser."
  },
  {
    "taskID": 1,
    "semanticTitleOfTask": "Subagent 4 - Input Handling for JSON Validation",
    "taskDescription": "This task receives the assembled JSON file from Subagent 3, prepares the data for the validation step, and ensures the input is recognized and ready for processing.",
    "inputDescription": "It requires the final assembled JSON from Subagent 3 in string format, representing the search results data that needs validation.",
    "inputRequired": [
      "assembled_json_file"
    ],
    "outputDescription": "Passes along the same JSON string for validation in the subsequent task.",
    "outputName": "json_to_validate",
    "promptInstruction": ""
  },
  {
    "taskID": 2,
    "semanticTitleOfTask": "JSON Structure Validation and Correction via LLM",
    "taskDescription": "Uses skill #223 to verify that the provided JSON is structurally valid and meets the specified format requirements, automatically correcting any issues found while preserving the integrity of the data.",
    "inputDescription": "The JSON string requiring validation, plus format requirements: a 'query' field and a 'results' array of objects with 'title' and 'url' fields, and an optional 'snippet'.",
    "inputRequired": [
      "json_to_validate"
    ],
    "outputDescription": "A validated JSON string that retains valid data, updates any issues, and returns the final structure required.",
    "outputName": "validated_json_file",
    "promptInstruction": "You are a JSON validation expert. I will provide you with a JSON string containing search results. Your task is to:\n1. Verify that the JSON is structurally valid\n2. Confirm it contains a 'query' field and a 'results' array of objects with 'title', 'url', and optionally 'snippet'\n3. Check for/fix any issues (missing fields, malformed syntax, invalid data types, duplicates, malformed URLs). If valid, return unchanged. If issues exist, fix them. For corrections, add 'validation_notes'. Maintain structure:\n{\n  \"query\": \"string\",\n  \"results\": [\n    {\n      \"title\": \"string\",\n      \"url\": \"string\",\n      \"snippet\": \"string\"\n    }\n  ],\n  \"validation_notes\": \"string\"\n}"
  },
  {
    "taskID": 1,
    "semanticTitleOfTask": "Merging multiple partial JSON inputs into one consolidated result",
    "taskDescription": "This task gathers and merges multiple partial JSON structures holding search results into one consolidated JSON. It doesn't require an LLM and is performed internally.",
    "inputDescription": "This task requires one or more JSON strings or objects containing partial search result data. They may come from preceding tasks or external sources, and need to be merged into a single uniform JSON structure.",
    "inputRequired": [
      "partial_json_1",
      "partial_json_2",
      "partial_json_n"
    ],
    "outputDescription": "It provides a unified JSON string that merges all provided partial search result objects, forming the basis for further data processing steps.",
    "outputName": "merged_json",
    "promptInstruction": ""
  },
  {
    "taskID": 2,
    "semanticTitleOfTask": "Removing duplicates from consolidated search results using an LLM-based approach",
    "taskDescription": "This task uses a powerful LLM to examine the consolidated JSON of search results, detect and remove duplicates based on matching title and URL, and produce a cleaned dataset without duplicate entries.",
    "inputDescription": "This task requires the merged JSON string containing search results and a set of deduplication instructions. The LLM will remove duplicate entries by comparing both title and URL for each entry.",
    "inputRequired": [
      "merged_json",
      "deduplication_instructions"
    ],
    "outputDescription": "A refined JSON string with no duplicate entries, preserving unique snippet information for merged items.",
    "outputName": "deduplicated_json",
    "promptInstruction": "You are tasked with analyzing search results and removing duplicates. Consider entries as duplicates if they have both matching titles AND URLs. When duplicates are found, merge them while preserving any unique information in their snippets. Analyze the following JSON search results and return a new JSON array with all duplicates removed. The output must maintain the exact same structure with `title`, `url`, and `snippet` fields for each entry. Ensure the output is valid JSON format."
  },
  {
    "taskID": 3,
    "semanticTitleOfTask": "Reordering cleaned search results using an LLM-based approach",
    "taskDescription": "This step uses an LLM to reorder search results based on specified sorting or ranking criteria, generating a final sorted dataset in valid JSON format.",
    "inputDescription": "This task requires the deduplicated JSON from the previous step, along with instructions on how to prioritize the ordering of each result entry. The LLM will produce a newly sorted JSON array.",
    "inputRequired": [
      "deduplicated_json",
      "sorting_instructions"
    ],
    "outputDescription": "A JSON array containing the same search results sorted according to the provided criteria, preserving valid structure.",
    "outputName": "sorted_json",
    "promptInstruction": "You are tasked with reordering search results according to specified criteria while maintaining valid JSON format. Sort the results based on the following priority: 1) Relevance to the original query 2) Completeness of information (entries with all fields filled should be prioritized) 3) Length and quality of snippets. The output must maintain the exact same structure with `title`, `url`, and `snippet` fields for each entry. Return the sorted results as a valid JSON array."
  },
  {
    "taskID": 4,
    "semanticTitleOfTask": "JSON Output Formatter for final search results",
    "taskDescription": "This is an internal data handling step that finalizes the JSON data from the previous step, ensuring it is well-formed and ready to be returned to the main workflow.",
    "inputDescription": "It requires either the sorted JSON (if sorting was performed) or the deduplicated JSON (if sorting was skipped), and ensures the final JSON is properly structured for handoff.",
    "inputRequired": [
      "sorted_json",
      "deduplicated_json"
    ],
    "outputDescription": "Outputs the final validated JSON string, ensuring correctness of structure and readiness for the main workflow.",
    "outputName": "final_json",
    "promptInstruction": ""
  },
  {
    "taskID": 224,
    "semanticTitleOfTask": "Web Search Query Execution",
    "taskDescription": "This task receives the user’s search query, connects to a web search engine, and returns raw textual search results with page titles, URLs, and snippet previews.",
    "inputDescription": "This requires VARIABLE1, the user’s search query in string format. It will be used to call an external web search API or engine, retrieving relevant search results from the open web.",
    "inputRequired": [
      "VARIABLE1"
    ],
    "outputDescription": "The output is a body of unstructured text containing multiple search results, each with a title, URL, and an optional snippet describing that webpage.",
    "outputName": "raw_search_results_2",
    "promptInstruction": ""
  },
  {
    "taskID": 223,
    "semanticTitleOfTask": "Search Results Structured Parsing",
    "taskDescription": "This step receives the raw text from the web search results and uses an LLM-based parser to systematically extract each search result’s title, complete URL, and optional snippet. If a snippet is unavailable, it is omitted.",
    "inputDescription": "This task requires the unstructured raw text from the previous step. The text may contain multiple entries, and the parser identifies each result’s pertinent fields (title, URL, snippet).",
    "inputRequired": [
      "raw_search_results_2"
    ],
    "outputDescription": "The output is a structured array of objects, each containing a 'title', 'url', and (if available) 'snippet'. This normalized format enables easy JSON assembly in the subsequent task.",
    "outputName": "parsed_search_results_2",
    "promptInstruction": "You are a search results parser. Take the following raw search results and extract each individual result into a structured format. For each result, identify and extract: 1) page title, 2) complete URL, 3) a brief descriptive snippet (if available). Format each result as a clean, structured object. Ensure URLs are complete and valid. Remove any advertising or sponsored content. If a snippet is not available, omit it from the object rather than leaving it empty."
  },
  {
    "taskID": 223,
    "semanticTitleOfTask": "Final JSON Structure Assembly",
    "taskDescription": "This step generates the final JSON output, combining the user’s original search query with the array of parsed search results in a valid JSON structure.",
    "inputDescription": "This task accepts two inputs: (1) the original user query string (VARIABLE1), and (2) the structured array of parsed results from the previous step (parsed_search_results_2). These elements are combined into a single JSON object.",
    "inputRequired": [
      "VARIABLE1",
      "parsed_search_results_2"
    ],
    "outputDescription": "The output is a complete JSON string containing the original query and a 'results' array of individual objects. It is prepared for saving as 'search_results.json'.",
    "outputName": "assembled_json_file_2",
    "promptInstruction": "You are a JSON formatter. Create a properly structured JSON object that combines: 1) the original search query, and 2) an array of parsed search results. Follow the schema: { \"query\": [original search query], \"results\": [array of objects] }. Ensure the output is valid JSON that can be parsed by standard JSON parsers. Each result object in the array should contain title, url, and optional snippet fields. Remove any malformed entries or invalid characters that would break JSON parsing."
  }
]

Happy? Now Copy-Paste To Proceed...

BACK TO SUBAGENTS

COPY TO CLIPBOARD