When to use them
Use Python or SQL for custom logic that the dedicated actions (Augment, Join, Extract Tables, and so on) don’t express directly:- Reshaping, pivoting, or aggregating columns
- Custom cleaning, parsing, or feature derivation
- Filtering and reformatting before embedding
- Calling external APIs from your API Secret Vault (Python)
On the canvas these appear as Python (
execute_python) and SQL (execute_sql) actions. The agent’s reasoning chain refers to the same execute_python capability when it runs code to answer a question.Python
The Python action runs an AI-assisted Python transform over the input table. Describe the transformation in plain language; Corvic drafts the code, and you can inspect and modify it before running. Python is also how agents call connected external APIs — credentials from the API Secret Vault are injected automatically, so your code references a connection by name rather than handling raw secrets.📸 Screenshot to add: the Python action panel (prompt, generated code, run) — save as
/images/action-python.pngSQL
The SQL action runs an AI-assisted SQL query over the input table. Use it for set-based transformations — joins, aggregations, window functions, and filtering — expressed in familiar SQL.📸 Screenshot to add: the SQL action panel (prompt, generated query, run) — save as
/images/action-sql.pngWhere they run
Python and SQL actions can be applied to Corvic Table outputs, and can also run directly on Space outputs — letting you post-process embeddings and their associated data without leaving the canvas.Usage
Review and edit
Inspect the generated code, adjust it as needed, and copy it if you want to reuse it.
Related Documentation
Workflows
Build pipelines on the canvas.
API Secret Vault
Call external APIs from Python.
Action Reference
See all available canvas actions.

