ORCFLO Documentation

Overview

When your workflow steps use tools like web search, web crawl, or content extraction, the raw tool results are collected into a data bucket. You can preview this structured data directly in the step detail panel and download it as a JSON file.

No additional credits are charged for downloading data. You already paid for the tool calls that produced the results.

What is the Data Bucket?

The data bucket is a JSON file that stores all raw tool results from a workflow step. Unlike the step output (which is the AI's synthesized response), the data bucket contains the unprocessed results returned by each tool invocation.

This is useful when you want to:

Inspect the raw data that the AI used to generate its response
Export tool results for use in external systems or analysis
Debug unexpected AI outputs by examining the source data
Archive structured research results from web scraping workflows

Accessing Step Data

After a workflow execution completes, click on any step to open the step detail panel. If the step has collected tool result data, you will see a Data section between the Output and Files sections.

Open the execution detail view for a completed workflow run
Click on a step that used tools (web search, crawl, etc.)
Look for the Data section with the database icon
Click to expand the section and view the JSON preview

The Data section loads content on demand. It won't fetch the data until you expand the section, keeping the UI fast even for large data sets.

Data Structure

The data bucket JSON is organized by tool name. Each tool has an array of results, one per invocation.

json

{
  "tools": {
    "tavily_crawl": [
      {
        "tool_use_id": "toolu_abc123",
        "result": {
          "url": "https://example.com",
          "content": "..."
        }
      }
    ],
    "web_search": [
      {
        "tool_use_id": "toolu_def456",
        "result": {
          "query": "AI trends 2026",
          "results": [...]
        }
      }
    ]
  },
  "result": {
    // Step-level structured result (if applicable)
  }
}

The tools object groups results by tool name, making it easy to find all results from a specific tool. The optional result field contains step-level structured output from transform steps or forEach accumulations.

Download & Copy

The Data section provides two ways to export the data:

Action	Description
Download JSON	Saves the full data bucket as a JSON file named `stepname_data.json`
Copy	Copies the full JSON content to your clipboard

For large data sets, the preview is truncated to keep the UI responsive. Click "Show full data" to see everything, or use Download JSON to get the complete file.

Which Steps Have Data?

The Data section only appears for steps that used tools and collected results. Common step types that produce data:

AI Model steps with tools enabled (web search, crawl, code interpreter, etc.)
Custom Agent steps that invoke tools during execution
ForEach loops that accumulate tool results across iterations