Python Data Analysis

Sandboxed Python execution for data analysis, charting, and statistical computation.

Atlas can execute Python code in a sandboxed environment for data analysis, statistical computation, and chart generation. The agent writes Python after running SQL queries to produce visualizations and deeper analysis.

Enable

ATLAS_PYTHON_ENABLED=true

Python execution requires a sandbox backend. Set ATLAS_SANDBOX_URL (sidecar) or use nsjail/Vercel sandbox. Unlike the explore tool, there is no just-bash fallback — Python will not work without a sandbox.

Sandbox Backends

Python runs through the same sandbox infrastructure as the explore tool, with the following priority:

Priority	Backend	How to enable
1	Sidecar	Set `ATLAS_SANDBOX_URL`
2	Vercel sandbox	Set `ATLAS_RUNTIME=vercel` (Python 3.13)
3	nsjail (explicit)	Set `ATLAS_SANDBOX=nsjail`
4	nsjail (auto-detect)	nsjail binary on `PATH` or `ATLAS_NSJAIL_PATH`

Unlike the explore tool, there is no just-bash fallback for Python. If no isolated backend is available, Python execution is rejected with an error.

Available Libraries

The following libraries are available in the sandbox:

pandas -- DataFrames and data manipulation
numpy -- Numerical computation
matplotlib -- Static charts and plots
plotly -- Interactive charts

The exact set of available libraries depends on the sandbox backend and its Python environment.

Chart Rendering

When Python code generates charts (via matplotlib or plotly), they appear inline in the chat UI. The agent can produce:

Bar charts, line charts, pie charts, scatter plots
Statistical visualizations (histograms, box plots)
Multi-series and grouped charts

Charts generated with Recharts format (a React charting library) are rendered natively in the web UI for interactive exploration.

Security Model

Python execution is isolated at multiple layers:

Import guard -- Defense-in-depth blocking of dangerous modules: subprocess, os, socket, requests, pickle, tempfile, shutil, and others. This is a first-pass filter; the sandbox backend is the real security boundary
No filesystem writes -- The sandbox environment is read-only
No network access -- Outbound connections are blocked by the sandbox
No shell access -- os.system(), subprocess.run(), and similar are blocked
Timeout enforcement -- Each execution has a configurable time limit

Environment Variables

Variable	Default	Description
`ATLAS_PYTHON_ENABLED`	—	Set to `true` to enable Python execution
`ATLAS_PYTHON_TIMEOUT`	—	Per-execution timeout

The sandbox itself is configured via the standard sandbox variables. See Environment Variables and Sandbox Architecture.

How It Works

The agent runs a SQL query via executeSQL and gets tabular data
The agent writes Python code using the query results as input (passed as a data parameter with columns and rows)
Atlas validates the code against blocked imports and builtins
The code runs in the sandbox backend with the data available
Output (text, tables, or charts) is returned to the agent and displayed in the chat UI

The agent decides when to use Python based on the question -- statistical analysis, trend detection, and visualization requests typically trigger Python execution.