> ## Documentation Index
> Fetch the complete documentation index at: https://docs.maadify.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Index Data Store
> Store internal data and make it available to agents
## What an index data store is
Index data stores hold internal knowledge that agents can search in chat or workflows. Each data store source defines where data comes from, how records are structured, which fields are searchable, and how users can add or import content.
The **Index Data Store** page has two main areas:
* **Data Store Sources** where you configure the sources that load data into index stores.
* **View Data** where you search, filter, import, add, edit, download, and delete indexed content for the selected source.
## Review data store sources
The **Data Store Sources** section shows each configured source as a card. Cards include the source name, description, connector, owner, and collection when available.
From each card, you can:
* Open the data source configuration.
* View the data loaded for that source.
* Delete a source if you have permission.
Use search and pagination to find a source when your tenant has many data stores.
## Add a data store source
You can create a source in two ways:
* **Add Data Store Source**: choose an existing action tool that returns data you want to index.
* **Add Import Store**: create a custom import store for manual file uploads and manually added content.
When you add a data store source from a tool, Maadify shows searchable tool cards. You can search tools, include eligible shared tools, and select the tool that should load data into the store.
Go to **Index Data Store**.
Select **Add Data Store Source** to use a tool, or select **Add Import Store** for manual imports.
Search for an available tool, review the connector and owner, and select the tool.
Fill in setup fields, choose the collection, select source and content fields, and configure any tool defaults.
Click **Add Configuration** or **Update Configuration**.
## Configure setup fields
The configuration dialog contains tabs that help you review and configure the source:
* **Setup Configurations**: choose the schema mode, collection, source column, and content columns.
* **Tool Configurations**: configure default values used when the selected tool loads data.
* **Details**: review the tool name, key, description, connector, and tenant owner.
* **Schema**: review the input schema and input example.
* **Output Example**: review a sample output when the tool provides one.
### Schema mode
The **Schema** setting controls which fields define the data store records.
* **Input** uses the selected tool's input schema.
* **Output** uses the selected tool's output schema.
* **Custom** lets you define structured fields with the schema builder.
Some sources only support specific schema modes. For example, import stores use **Custom** because you define the shape of manually imported content.
### Collection
The **Collection** setting controls where the data is stored. Available collections include:
* **product**
* **order**
* **inventory**
* **knowledge\_base**
* **shipment**
Choose the collection that matches how users and agents will search the data.
### Source column
The **Source Column** identifies the source record or file. Maadify uses it for display, deduplication, and grouping. For file-based import stores, this is commonly `_file_name`.
### Content columns
**Content Columns** define which fields are embedded for search. Choose fields that contain meaningful searchable text, such as descriptions, notes, titles, messages, or document content.
Use stable IDs for the source column and rich text fields for content columns. This makes search results easier to trace back to the original record.
## Create custom structured inputs
When you choose **Custom** schema mode, use the schema builder to create structured inputs for records in the index data store. These fields become the form users fill out when they add or edit content.
You can create fields manually or generate them from a JSON schema or example JSON response.
The schema builder supports:
* **Field names and display names** for each value.
* **Types** such as string, integer, number, boolean, object, arrays, and object arrays.
* **Required fields** for values that must be provided.
* **Descriptions** that explain what each field means.
* **Options** for fields that should use a fixed list of values.
* **Defaults, ranges, and length limits** for supported field types.
* **Nested object properties** for structured records.
As you add fields, Maadify generates the schema used by the data entry form and updates the available source and content column choices.
## Configure tool defaults
For tool-backed data store sources, the **Tool Configurations** tab controls how Maadify calls the selected tool when importing data.
You can either:
* Use the tool owner's default configuration.
* Provide your own default values for the tool.
These values are used when you import data from the source. They help keep recurring imports consistent.
## Import data
After you configure a data store source, select it and use **Import Data**.
Depending on the source, you can:
* Run the configured tool to load data into the selected collection.
* Enable **Manual Upload** and upload a file.
* Process uploaded files into the index store.
Manual upload is useful when you have files that should be indexed directly. Tool import is useful when data should come from a connected application.
## View and manage data
After you select a configured source, the **View Data** section shows indexed records for that source.
You can:
* Search across indexed content.
* Filter by available metadata columns.
* Toggle **Show Full Document** to switch between full documents and smaller indexed nodes.
* Open a record to view or edit it.
* View files linked to a record.
* Download record data.
* Delete records when you have permission.
Use the search box and filters to find records.
Select a row or use the row actions menu.
Edit structured fields when the store has a schema, or edit text content for simple/manual stores.
Click **Add** or **Update** when editing is available.
## Use data stores in chat and agents
After data is indexed, users can search it with **Universal Search** in chat. Parent agents can also use configured data store tools during workflows when they need internal knowledge.
Keep schemas focused. Choose content columns that improve search quality, and avoid storing sensitive data unless agents need it for the workflow.