Runtime CSV transforms agent
Use an AI agent to reshape CSV lists to match your template's requirements, re-run fresh on every file.
Runtime CSV transforms agent reshapes CSV lists to match a target template's requirements, similar to CSV transforms agent. The difference is timing: instead of generating and caching a single code snippet the first time, this transform runs the agent fresh on every file at runtime. That makes it more resilient to input shapes that change from run to run, at the cost of running the agent more often.
Availability: Beta. Requires the
WorkflowsRuntimeCsvAgentfeature. Contact your OneSchema support representative if you do not see this transform.
What it does
- Input: one or more CSV lists.
- Output: one or more CSV lists, reshaped to match your target template's requirements.
- Rather than relying on a single previously generated code snippet, the agent evaluates each file at the time it runs and produces the transformation on the spot.
When to use it
- Your source file's column layout, order, or naming changes from run to run — for example, a vendor that reorders columns each month.
- A cached code snippet from CSV transforms agent has started failing on newer files with a different shape.
- You'd rather trade a small amount of extra processing time for more resilience to layout drift.
How to configure it
Add a Runtime CSV transforms agent node after the node that produces the CSV list you want to reshape, and after your target template is configured. Provide instructions describing the mapping directly in the node panel.
Settings
| Setting | Description | Default |
|---|---|---|
| Instructions | The plain-language description of how the output should be shaped, provided directly in the node panel. There is no cached code to edit — the agent transforms fresh on every run. | Empty |
Example
Input
A monthly vendor CSV where column order shuffles from month to month.
Output
A CSV consistently matching your template's columns, regardless of the source column order that month.
Troubleshooting
Runs take longer than expected
- This is expected — the agent evaluates each file at runtime rather than reusing cached code. If speed matters more than layout resilience, consider CSV transforms agent instead.
Output is inconsistent between runs
- Tighten your instructions and your template's column descriptions so there's less room for the agent to interpret ambiguous cases differently.
See also
Updated about 7 hours ago