On this page
Kater CLI Explore and try ideas explore definition
kater explore definition generate
Saves a shared model's Explore file, with its questions, and with --llm plans that an AI model makes.
Use it to make the file, or to plan the model’s questions. It needs write access to the file, beside the model.
Examples
# Save the Explore file of the ad_reporting model again, while it has this revisionkater explore definition generate --model semantic.ad_reporting \ --scope connections/snowflake_prod_connection/modeling \ --expected-revision 546c87e5d0d229965dc939b1d7b152aac6762c15e61884fd20b75ba2c7b1b37a# Also plan up to two questions with an AI model, and add new questionskater explore definition generate --model semantic.ad_reporting \ --scope connections/snowflake_prod_connection/modeling \ --expected-revision 546c87e5d0d229965dc939b1d7b152aac6762c15e61884fd20b75ba2c7b1b37a --llm# Plan one question, by the id that kater explore use-case list giveskater explore definition generate --model semantic.ad_reporting \ --scope connections/snowflake_prod_connection/modeling \ --expected-revision 546c87e5d0d229965dc939b1d7b152aac6762c15e61884fd20b75ba2c7b1b37a \ --question 8011ea1a161a2829d33e04820aa3cca358c5e553ad475f2f6b1caeeace9213d1Options
It also takes the global options --json, --kater-root, --scope and --help.
What it saves
- Without
--llm. The file holds the questions the model’s authors wrote. An existing file keeps its questions and plans. - With
--llmor--question. An AI model also plans questions, and with--llmalone it adds new ones. - A model that must be built first. It saves nothing, and says why.
- A hosted Kater. Planning needs your organisation’s AI provider and its API key.
In a request file, "dry_run": true checks the save and writes nothing. A dry run cannot plan questions.