> For the complete documentation index, see [llms.txt](https://docs.dropzone.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.dropzone.ai/integrations/data.md).

# Data Source Integrations

These integrations are Data Source.

Data Sources enrich the information Dropzone AI uses to perform alert investigations and respond to interactive chat. Dropzone has support for many Threat Intelligence (TI) feeds, tools, and corporate systems such as identity, directory, and SIEM tools.

For more details about Data sources see the [Data Sources definition](https://docs.dropzone.ai/dropzone-101/terms-and-defs/data-sources) page.

## Query Guidance

Query guidance is a Data source feature that can be enabled per-integration. It is designed to assist Dropzone’s AI SOC Agent in locating and retrieving relevant information during investigations, reducing response times and improving resource usage.

## How does it work?

Query guidance entries have two parts: a **Condition** and an **Instruction**.

**Conditions** decide in what scenarios the guidance should be applied. This should be a description of the kind of question the AI might be needing to answer.

**Instructions** tell the AI what to do when that condition matches. These should be concrete steps and instructions that will be loaded verbatim into the AI’s context. You may use numbered steps or line breaks to order them.

Every condition is shown to the AI at the start of a query. When a condition fits the question it is working on, the AI loads that instruction before it writes the query.

Example:

| Condition                                                   | Instruction                                                                                                                                                              |
| ----------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| Questions about user sign-in or authentication activity     | Look for sign-in events in `okta_systemlog`. Treat a sign-in as successful when `outcome` is `success`. Aggregate by application and source IP when building a baseline. |
| Questions about process or executable launches on endpoints | Look for process launches in `winlogbeat`. Filter by the host name field, not the host ID.                                                                               |

## Adding Query Guidance

Query Guidance is currently enabled on the following integrations:

To add query guidance, do the following:

* Open the Data source integration you want to access
* Navigate to the Query Guidance section, then click “Enable”
* Input the Condition/Instruction pairs individually, clicking “Add Item” for each

<figure><img src="https://3968177484-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FDt9gEGwbp9ApiQAJP3hQ%2Fuploads%2Flw5YSvFMWmUDeJhrexVG%2Fui-query-guidance.png?alt=media" alt=""><figcaption><p>Example: Add Query Guidance</p></figcaption></figure>

## Best Practices

Conditions should be specific enough to recognize and not too narrow to exclude use.

* “Questions about user activity” is too broad and will be applied too often
* “The user bulk downloaded files” is too narrow
* “Questions about user download activity” is specific and broad enough

Instructions should be detailed enough to follow without guessing. Vague instructions can be followed literally and still produce weak information.

* “Look in Splunk indexes for organizational data” is too vague
* “Look for download activity in the Splunk index `_audit_user_activity`” is detailed enough
