August 20, 2026: Cortex Agent code execution tool (Preview)¶

The Cortex Agent code execution tool is now in public preview. When you enable the tool on an agent, the agent can generate and run Python in a secure, isolated sandbox to process data, perform calculations, and produce visualizations. The sandbox includes the Python standard library along with preinstalled data-processing and plotting libraries such as numpy, pandas, scipy, pyarrow, matplotlib, and plotly. You can install additional PyPI packages through the Artifact Repository.

Note the following behavior:

• The sandbox doesn’t query your data. When the agent needs data from Snowflake, it runs the query with its SQL tools outside the sandbox, and the sandbox works with the results.
• A sandbox is scoped to a single conversation thread. Files written to the mounted workspace persist because the workspace is backed by a stage, but in-memory state doesn’t carry over between separate code executions.
• Calls to agents that use owner’s rights don’t support code execution. Invoke the agent with caller’s rights if it needs the tool.

For more information, see Cortex Agent code execution tool.