kater app ask

Asks the person where a figure on an app should come from, when no model in Kater has its data.

An agent uses it only after it searched the workspaces, and then the shared models. It answers with the question and four fixed options, and writes nothing.

Examples

# Ask where the monthly spend budget on the paid_media_performance app should come from
kater app ask connections/snowflake_prod_connection/workspaces/ad_reporting/apps/paid_media_performance --request-file spend_budget_ask.json
# Ask the same for a draft app in your brainstorm
kater app ask connections/snowflake_prod_connection/users/you@example.com/brainstorm/platform_spend/apps/platform_spend_review --request-file spend_budget_ask.json

Arguments

Argument	Description
`<app>`	The app the figure is for, by its folder or one of its pages. It can be an app in your own folder, an app in a workspace, or a draft app in your brainstorm.

Options

Option	Description
`--request-file <file>`	Required. The figure, what Kater does not have, and what the real data must hold, as JSON. Give `-` to read it from standard input. See The request.

It also takes the global options --json, --kater-root, --scope and --help.

The request

Field	What it holds
`figure`	The figure as the page labels it, such as “Monthly spend budget”. It has 120 characters or fewer.
`missing`	What Kater does not have, as words that follow “I could not find”, such as “media budgets”.
`searched`	What you searched in Kater, as words that follow “I searched”, such as “the ad_reporting workspace, and the Ad Reporting model”.
`needs`	The data that would take the place of mock data: the `measure` it counts, its `grain` (one row of what), and its `period` (which rows). It can also hold its `columns` and how the figure `computes` from them.
`ask`	The one question for the data team.
`reason`	Optional. Why the page has no real figure. Without it, Kater makes one from `missing`, such as “No media budgets in Kater yet.”
`add`	Optional. How the data gets into Kater, such as “Model the budget table in the Ad Reporting model”.
`id`	Optional. The source’s id on the page. Without it, Kater makes one from the figure.

{"figure": "Monthly spend budget", "missing": "media budgets", "searched": "the ad_reporting workspace, and the Ad Reporting model", "needs": {"measure": "Planned spend", "grain": "One row a month for each platform", "period": "The last 12 complete months"}, "ask": "Is the media plan's budget in Snowflake? If it is, can it go into the Ad Reporting model?"}

The four options

The answer holds the question, and four options in a fixed order. Each option says what the page gets for that answer: a <kater-source> with its <kater-cite>, or a Kater tag. The parts that the answer gives are in braces.

Add it to Kater. When the connection is not ready, this option is Finish connecting Kater.
I will attach a file.
Give a link to another system.
Use mock data. The figure then shows a Mock data badge.

Use no other source until the person answers. Through the Kater MCP, the agent asks the person in the chat’s own question form, where the chat has one.