Skip to main content

Documentation Index

Fetch the complete documentation index at: https://datost.com/docs/llms.txt

Use this file to discover all available pages before exploring further.

Notebooks are currently feature-flagged (NOTEBOOK_MODE). Availability depends on your plan — reach out to your Datost contact if you’d like it enabled for your workspace.
Datost Notebooks give your team a full Python environment that knows about your data. Each notebook runs in an isolated E2B sandbox, so heavy analysis, ad-hoc scripting, and exploratory work stay cleanly separated from your warehouse and from other notebooks.

​
What you get

  • A real Python kernel — not a sandboxed subset. Install packages, import anything, and iterate on cells just like Jupyter.
  • Direct access to your connected data — bind a data source to a notebook and Datost auto-injects a datost Python client you can use with from datost import query.
  • Two editing modes — Datost’s native cell UI, or a full embedded JupyterLab session for power users.
  • Persistent notebooks, ephemeral sessions — your cells, outputs, and generated files are saved; the compute sandbox spins down after 30 minutes of inactivity and resumes on demand.

​
Creating a notebook

1

Open the Notebooks page

Navigate to Notebooks in the Datost sidebar and click New Notebook.
2

Bind a data source (optional)

From the header, pick any healthy connected warehouse. Once bound, Datost writes a datost.py client into the sandbox and pre-installs requests and pandas.
3

Add and run cells

Add code or markdown cells, drag to reorder, and hit run. Each execution streams stdout/stderr, rich outputs (PNG, JPEG, SVG, HTML, text), and errors back into the cell.
4

Switch to JupyterLab (optional)

Toggle JupyterLab in the header to open the same notebook inside a full embedded Jupyter server backed by the same sandbox.

​
Querying your data

With a data source bound, run SQL straight from Python: 
from datost import query

df = query("""
  select date, revenue
  from core.daily_revenue
  where date >= current_date - 30
""")
df.head()
Results come back as a pandas DataFrame — chart them with matplotlib, slice them with pandas, or hand them to any library you pip install.

​
Runtime and libraries

Pre-installed at session start: pandas, requests, jupyterlab, ipykernel. Install anything else with !pip install <package> in a cell, or use the install packages action.
The runtime is standard CPython 3 running inside E2B’s Code Interpreter. Sessions auto-timeout after 30 minutes of inactivity and can be manually ended from the header.

​
Outputs, files, and sharing

  • Rich outputs (tables, charts, HTML) are rendered inline and stored on each cell so they persist between sessions.
  • Generated files written to the sandbox are tracked in notebookOutputFiles and available for download.
  • Notebook state (cells, order, outputs, bound data source) is saved automatically — reopen a notebook any time and pick up where you left off.
The sandbox filesystem is ephemeral. Anything not saved as a notebook output file will be lost when the session ends. For long-running results, persist them to your warehouse or download them.
Notebooks are scoped to your organization, so any teammate with access to your Datost workspace can open, run, and edit them.