> For the complete documentation index, see [llms.txt](https://v2.dataos.info/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://v2.dataos.info/build/v1/understand/explore-data/workbench.md).
# Workbench
Once the metadata looks promising, validate the data itself. Workbench connects to a source, runs SQL, views results, and lets you browse schemas and tables from one workspace. Use it before you commit a dataset to a model, a report, or an application.
## What is in Workbench
| Area | What it is for |
| --------------- | ------------------------------------------------------------------ |
| Workbench home | Create a new bench, open a saved one, or pick up a draft |
| New bench setup | Choose the data source and the Secret to connect with |
| Query workspace | Write SQL, run queries, browse schemas and tables, inspect results |
## 1. Open Workbench and start a bench
Open **Workbench** from the left navigation. The home page shows your saved benches. From here you can:
* Create a new bench.
* Reopen unsaved benches.
* Return to a previous one.
Click **New bench** to start.
## 2. Pick a source and a Secret
Choose the data source (Depot) from the list (filter by type, owner, or date). Then choose the **Secret** Workbench connects with.
Workbench needs both. A single Depot can have several Secrets, each representing a different permission context, so it cannot safely pick one for you. Choose the Secret that matches what you intend to do. These purposes are defined for the secrets, while registering in [Depot](/build/v1/understand/connect-sources.md).
| Purpose | Use it when |
| -------- | ----------------------------------------------------- |
| `query` | You only need to query and explore |
| `scan` | You need metadata scanning or schema discovery |
| `rw` | You need read-write access and have permission for it |
| `full` | the credential has complete access to the data source |
| `masked` | the credentials masks some of the data |
{% hint style="info" %}
When you share a bench, the selected Secret sets the permission context others inherit. As a bench owner, you can choose a masked, read-only Secret so others explore without seeing sensitive fields.
{% endhint %}
## 3. Browse, then query
With the source and Secret selected, the bench opens. The left panel shows the source's schemas, tables, and columns. Expand a schema, select a table, and read its columns and types. Look for identifier fields, categorical fields, possible measures, and fields that will need filtering or transformation. Decide what is relevant before you write SQL.
Type your query into the editor and run a cell, or **Run all** for every cell. The execution time and returned rows tell you whether the query did what you expected.
## 4. Refine with cell references
After the first result, iterate. You can run the next query against the source, or run it locally against the previous cell's output using a cell reference. Cell references skip the round trip to the source, so they are faster and cheaper. Add a cell and click the cell-reference icon to enable it.
{% hint style="info" %}
Cell references run locally. Performance scales with dataset size and your machine's memory.
{% endhint %}
## 5. Work across benches and save
Each bench is a tab. Open several to compare sources or tables, or to keep drafts separate from validated queries. Unsaved benches stay around until you close or save them. Save a bench when you want to continue later, prepare a query for downstream work, or share the exploration.
## What is next
| Outcome | Next action |
| --------------------- | ----------------------------------------------------------------------------------------------- |
| Keep exploring | Inspect more tables, columns, or values |
| Use it in a build | The dataset supports a reusable use case. Go to [Productize](/build/v1/productize/overview.md). |
| Try another source | The selected source does not have the data you need |
| Source the data first | The data is not yet reachable by your engine. See [Move](/build/v1/move/overview.md). |
---
# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.
## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.
Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:
```
GET https://v2.dataos.info/build/v1/understand/explore-data/workbench.md?ask=&goal=
```
`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.
Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.