skippr ask
skippr ask runs a read-only data-engineering question against the configured pipeline: it uses the same authenticated, hosted runtime as skippr model, but does not apply patches or change warehouse state beyond what a normal read-only SQL path would do.
Usage
bash
skippr [--config <path>] [--log [level]] ask --pipeline <name> --question "<text>" [--output text|json|jsonl]Flags
| Flag | Description |
|---|---|
--pipeline <name> | Pipeline to scope catalog and warehouse context. |
--question <text> | Natural-language question for the agent. |
--output text | Default terminal-oriented output. |
--output json / jsonl | Structured events for tooling. |
Global flags: --config, --log.
Prerequisites
Same authentication and balance expectations as skippr model. The pipeline must be fully configured in skippr.yml (source, sink, etc.) so the data-engineer suite can resolve providers.
See also
skippr plan— produce a plan without applying changes.skippr model— full modeling workflow with writes.