dx

name	dx
description	Run DX Data Studio SQL queries, saved query datafeeds, and CSV downloads
argument-hint	<operation> [args]
allowed-tools	Bash, Read

DX Skill

Use the local DX client for Data Studio query execution, query-run polling, saved query datafeeds, and result downloads: $ARGUMENTS

Configuration

Required in .env:

DX_API_TOKEN=...

Optional defaults for DX Cloud:

DX_WEB_BASE_URL=https://app.getdx.com
DX_API_BASE_URL=https://api.getdx.com

For dedicated or managed DX deployments, set the deployment-specific web and API base URLs.

Commands

Check Auth

python3 -m sidekick.clients.dx whoami

Use this first when diagnosing token or scope issues.

Run SQL

python3 -m sidekick.clients.dx query "SELECT id, name FROM github_repositories LIMIT 10"

The client submits the query, polls until completion, and prints a compact table. Add --json for the raw JSON response.

Run SQL From A File

python3 -m sidekick.clients.dx query-file path/to/query.sql

Template Variables

Use variables without the leading $:

python3 -m sidekick.clients.dx query \
  "SELECT * FROM github_repositories WHERE id IN ($repo_ids)" \
  --var repo_ids=1,2,3

For complex values:

python3 -m sidekick.clients.dx query-file query.sql \
  --variables-json '{"repo_ids":["1","2","3"]}'

Download Full Results

JSON results are limited by DX. Use CSV downloads for the full result set:

python3 -m sidekick.clients.dx query \
  "SELECT * FROM github_pulls" \
  --csv memory/dx-github-pulls.csv

For an existing query run:

python3 -m sidekick.clients.dx results QUERY_RUN_ID --csv memory/dx-results.csv

Query Run Status And Results

Submit without waiting:

python3 -m sidekick.clients.dx execute "SELECT 1"

Check or wait for status:

python3 -m sidekick.clients.dx status QUERY_RUN_ID
python3 -m sidekick.clients.dx status QUERY_RUN_ID --wait

Fetch JSON rows for a completed run:

python3 -m sidekick.clients.dx results QUERY_RUN_ID

Saved Query Datafeeds

Use a saved query datafeed token:

python3 -m sidekick.clients.dx datafeed FEED_TOKEN
python3 -m sidekick.clients.dx datafeed FEED_TOKEN --var team_id=123 --columns date,count

Save returned datafeed rows as CSV:

python3 -m sidekick.clients.dx datafeed FEED_TOKEN --csv memory/dx-datafeed.csv

DX AI Prompt-To-SQL

DX Data Studio AI can generate SQL from prompts in the DX product UI. As of the current public docs and DX CLI package, there is not a documented Web API or CLI command for submitting a natural-language prompt to the platform AI and receiving generated SQL.

When a user asks for platform AI query generation:

1. Use the DX UI at DX_WEB_BASE_URL and Data Studio AI when browser access is appropriate.
2. Paste the generated SQL into python3 -m sidekick.clients.dx query ... or save it to a .sql file and use query-file.
3. Do not call undocumented internal app endpoints unless the user explicitly asks for that brittle workflow.

The local command python3 -m sidekick.clients.dx ai "prompt" explains this limitation and exits without making an API request.

Output

By default, commands print compact tables for humans. Add --json for raw API payloads when another tool needs to parse the response. Use --csv path/to/file.csv when the result set may exceed JSON row limits.