Knowledge and Data Processing Nodes

4. Knowledge and Data Processing Nodes​

The Knowledge category contains the nodes that read, transform, store and query data in workflows. They are listed under the KNOWLEDGE heading in the Nodes tab of the node catalog, and it is the largest category.

Most of these nodes require no credential connection; they run on SkyStudio's own data infrastructure.

Nodes​

Which Node Should You Use?​

Table Operations​

File to Table​

Converts an uploaded file (Excel, CSV) into a database table, making the data queryable and reusable.

Parameters​

Example Use​

File Input → File to Table → Query → Text Output

File to Table settings:

• File: Output of the File Input node
• Table Name: satislar_2026

Note: Do not use spaces or non-ASCII characters in table names; separate words with underscores.

JSON to Table​

Converts JSON data into a table. Used to make API responses queryable.

Parameters​

Example Use​

HTTP Call → JSON to Table → Data Analysis

Using Root Path: if the API response looks like { "data": { "items": [...] } }, enter data.items in the Root Path field.

List Tables​

Lists all tables in your database. Used to find out which tables exist and what they are called.

Parameters​

Example Output​

{
  "tables": [
    { "name": "satislar_2026", "row_count": 1248, "created_at": "2026-08-01" },
    { "name": "musteriler", "row_count": 342, "created_at": "2026-07-15" }
  ]
}

Table Data​

Retrieves data from a given table. Required to use tables you created earlier within a workflow.

Parameters​

Note: Use the Limit parameter on large tables; loading an entire table into memory slows the workflow down.

Query​

Runs a filtered query against a collection or table. Unlike the Table Data node, it can apply conditional filtering and grouping.

Parameters​

Example Use​

• Collection: satislar_2026
• Query: tutar > 10000 AND bolge = 'Marmara'
• Sort: tutar DESC
• Limit: 20

Knowledge Base​

Knowledge Save​

Saves a file to the workspace's knowledge base. Saved documents can be used as references by your agents.

Parameters​

Example Use​

A workflow that automatically adds incoming mail attachments to the knowledge base:

Search Mail (Outlook) → Knowledge Save → Send Message (Slack)

Note: Documents added to the knowledge base can also be managed from the agent's Knowledge section described in Section 04.

Knowledge Search​

Runs an AI-assisted search over the workspace's knowledge base. It returns not only the matching text but an answer to your question together with the source document.

Parameters​

Example Use​

A chatbot that answers internal policy questions:

Text Input → Knowledge Search → Anthropic API Call → Text Output

Knowledge Search settings:

• Query: Output of the Text Input node, selected with /

Example Output​

{
  "answer": "Yıllık izin hakkı 1-5 yıl kıdem için 14 gündür.",
  "sources": [
    { "document": "IK-Politikalari-2026.pdf", "page": 12, "score": 0.94 }
  ]
}

Note: This node is the basis of internal question-and-answer workflows. It returns nothing while the knowledge base is empty — add documents first with Knowledge Save or from the Settings → Knowledge Base screen.

Data Transformation​

Text2data​

Extracts structured data from free text. Turns plain text such as an email body, a form response or a note into field-value pairs.

Parameters​

Schema Example​

{
  "firma_adi": "string",
  "fatura_no": "string",
  "tutar": "number",
  "tarih": "date"
}

Example Use​

A workflow that extracts the details from an incoming invoice email and records them in accounting:

Search Mail (Outlook) → Text2data → Create Purchase Invoice (Paraşüt)

Note: Use explicit field names in the schema; vague names (such as x or value) reduce extraction accuracy.

Transform​

Converts data from one format to another. Used to rename, combine or restructure fields.

Parameters​

Example Use​

To map column names in Excel to CRM field names:

When to Use It​

• When field names do not match between two nodes
• When a date or number format needs to be changed
• When several fields need to be combined into one

Formula​

Performs calculations and formula operations. Used for sums, ratios, percentages and conditional calculations.

Parameters​

Formula Examples​

Parameter​

Defines a fixed variable for use within the workflow. Lets you manage a value from a single place when it is used in more than one node.

Parameters​

Example Use​

To manage a Jira project key used in many places in the workflow from a single point:

• Parameter Name: jira_proje
• Value: ENG

In later nodes you select this parameter with /. When the project changes, updating this one node is enough.

Save Data​

Stores data produced in the workflow persistently, making it available in later runs.

Parameters​

Example Use​

Used in a workflow that runs daily to store the previous day's result and compare against it.

SkyData​

Accesses SkyStudio's data service. Used to query the data sources defined on the platform.

Parameters​

Form Builder​

Generates a form to collect structured input from the user. Used when you need specific fields to be filled in rather than free text.

Parameters​

Field Types​

Common Errors​

Notes​

• Most nodes in this category require no credential connection; they run on SkyStudio's own data infrastructure.
• Table names are used as references throughout the workflow; renaming one later affects every node that uses that table.
• Table Data retrieves the entire table while Query retrieves only the matching records; Query is faster on large data sets.
• A Knowledge Search answer is limited to the content of the knowledge base; the node produces no result while the knowledge base is empty.
• Do not confuse Text2data with Transform: Text2data extracts fields from free text, while Transform reshapes data that is already structured.
• Parameter suits constant values that recur throughout a workflow (a project key, a folder name, an email address); they are updated from a single place.
• Save Data and SkyData are different: Save Data stores the data the workflow itself produces, while SkyData reads the data sources defined on the platform.
• Form Builder is for collecting specific fields from the user instead of free text; if the field order changes, the output order changes with it.
• Give your nodes meaningful names; it makes finding the right output in the / menu easier.

Loop​

The Loop node enables you to iterate through an array or a numerical range and invoke a separate workflow for each element. It is designed for scenarios that require processing large numbers of records—such as hundreds of orders, customers, or data rows—individually through a sub-workflow.

The node enqueues each element for processing and immediately returns a unique batch identifier (batchId). All operations are executed asynchronously in the background, allowing the main workflow to continue without waiting for the batch to complete. You can monitor execution progress and retrieve results at any time using the batchId.

The Loop node is available in the node catalog under Nodes → Information.

Connection Setup and Management​

The Loop node uses an invocation token associated with the target workflow to execute a sub-workflow for each element. This token identifies the sub-workflow to be invoked.

Prerequisites​

Before configuring the Loop node, ensure that the target sub-workflow has been published and that its invocation token is available. You can obtain this token from the target workflow's trigger settings.

Note: The invocation token determines which sub-workflow will be executed. Configure separate connections for loops that invoke different sub-workflows. Using descriptive connection names, such as "Order Processing Loop" or "Reporting Loop," helps prevent configuration errors.

Adding a Connection in SkyStudio​

1. Click the chain icon (Application Connections) in the bottom toolbar of the workflow canvas.
2. Click + New Connection.
3. Select Loop from the provider list. It is located under the Other category. Alternatively, enter loop in the search field.
4. Complete the following fields:

5. Click Add Connection to save the configuration.

Note: Instead of completing each field individually, you can select Edit as Raw JSON and paste the connection configuration in JSON format.

Assigning a Connection to the Node​

After adding the Loop node to the workflow canvas, select the previously configured connection from the Application Connections dropdown menu in the right-hand configuration panel.

Alternatively, expand the Connection section and enter the Name and Workflow Token (Bearer) values directly in the node configuration.

Although direct configuration is convenient for testing, using a saved connection is recommended for persistent and production environments.

Note: The Queue Status operation does not require an authentication connection. A connection is required only for the two Enqueue operations.

Operations​

The Loop node provides three operations:

Enqueue (Array)​

This operation iterates through an array, enqueues a processing task for each element, and immediately returns a batchId.

Each array element is passed to the target sub-workflow through the query field.

Parameters​

Note: The Items and Payload fields support dynamic parameter references. Click either field and press / to select an output from a previous node.

Data Passed to the Sub-Workflow​

For each element, the following request body is sent to the target sub-workflow:

{ "query": <current_element>, ...payload }

Usage Example​

Items:

[
  { "id": "ORD-001", "musteri": "Eti Gıda", "tutar": 12500 },
  { "id": "ORD-002", "musteri": "Ülker", "tutar": 8300 }
]

Payload:

{ "kaynak": "loop" }

Example Output​

{
  "batchId": "batch-ab12-cd34",
  "enqueued": 2,
  "status": "queued"
}

Note: All iteration-specific data must be provided through the query field. The Payload parameter is intended exclusively for static contextual information. Data that varies between iterations should be included in the individual array elements rather than in Payload, allowing the sub-workflow to access it through query.

Enqueue (Number Range)​

This operation iterates through a numerical range defined by a starting value, an ending value, and an increment.

For each number in the range, the target sub-workflow is invoked. This operation is particularly useful for pagination and workflows that require a predetermined number of repetitions.

Parameters​

Usage Example​

When Start is set to 1, End to 5, and Step to 1, the following request bodies are generated for the target sub-workflow:

{ "query": 1 }
{ "query": 2 }
{ "query": 3 }
{ "query": 4 }
{ "query": 5 }

Pagination Example​

To process an API containing 100 pages, configure the operation as follows:

The target sub-workflow uses the incoming query value as the page number.

Note: When processing large numerical ranges, configure the Max Items limit carefully. If the range exceeds the configured limit, the remaining elements are not processed, and the output includes truncated: true.

Queue Status​

The Queue Status operation retrieves the execution progress of a batch identified by its batchId. It can also return individual execution results, including errors, when requested.

This operation is used to monitor batch completion and retrieve processing results.

No authentication connection is required for this operation.

Parameters​

Example Output​

{
  "batchId": "batch-ab12-cd34",
  "total": 5,
  "counts": { "done": 4, "failed": 1 },
  "complete": true,
  "results": [
    { "index": 0, "item": { "id": "ORD-001" }, "ok": true, "result": { "ok": true } },
    { "index": 1, "item": { "id": "ORD-002" }, "ok": false, "error": "..." }
  ]
}

Output Fields​

Note: You can periodically invoke this operation until complete: true is returned. If the target sub-workflows persist their execution results independently, retrieving and aggregating those results through Queue Status may not be necessary.

Reliability​

Note: The total execution time of a batch depends on the processing time required by the target sub-workflow for each element. The Loop node executes these operations asynchronously, without blocking the main workflow.

Quick Start Examples​

Example 1: Processing Orders Individually​

The following workflow processes order records from a table by invoking a sub-workflow for each order:

Text Input → Table Data → Enqueue (Array) → Text Output

Enqueue (Array) configuration:

• Items: Use / to select the output of the Table Data node.
• Payload: { "kaynak": "gunluk-siparis" }
• Max Items: 500

Example 2: Processing a Paginated API​

The following workflow processes an API containing 100 pages:

Webhook → Enqueue (Number Range) → Text Output

Enqueue (Number Range) configuration:

• Start: 1
• End (Inclusive): 100
• Step: 1

The target sub-workflow uses the incoming query value as the page number and performs the corresponding HTTP Request.

Example 3: Monitoring Batch Completion and Reporting Results​

The following workflow starts a batch operation, monitors its completion, and sends a notification when processing finishes:

Enqueue (Array) → Queue Status → If-Else Condition ─┬─ (True)  → Send Message (Slack)
                                                    └─ (False) → Queue Status

Queue Status configuration:

• Batch ID: Use / to select the batchId field from the Enqueue operation's output.
• Include Results: Enabled

If-Else configuration:

• Conditions: Use / to select the complete field.
• Operator: Equals
• Value: true

Example 4: Reporting Failed Records​

The following workflow identifies failed operations after batch completion and generates a summary report:

Queue Status → LLM (Summarize Errors) → Send Email (Outlook)

Queue Status configuration:

• Include Results: Enabled

The LLM node analyzes the results array and summarizes records where ok: false.

Troubleshooting​

Additional Notes​

• The Loop node is available in the node catalog under Nodes → Information, while its connection provider is located under the Other category.
• Enqueue operations return a batchId immediately, allowing the main workflow to continue while processing takes place asynchronously.
• The Queue Status operation does not require an authentication connection.
• Data that changes between iterations is always passed through the query field, while Payload remains constant throughout the loop.
• The End (Inclusive) value is included in the numerical range. For example, a range from 1 to 5 produces five iterations.
• The default Max Items limit is 10000. It is recommended to configure an appropriate processing limit for production environments.
• Define separate Loop connections for different target sub-workflows. Each connection is associated with a single target workflow.
• To monitor batch completion, combine the Queue Status operation with an If-Else Condition.