Command reference

Every command accepts --project <name> (or account/project), --account <name> and --json. Names can be the name, the short name shown by rd projects list, or an id; where a name is ambiguous the CLI asks you to be more specific. Run rd help <command> for details.

You and your projects

Command	What it does
`rd whoami`	Who you are and the projects you can use
`rd accounts list`	Your accounts
`rd projects list`	Your projects (● marks the default)
`rd projects use <account>/<project>`	Remember the account and project for the next commands
`rd auth login` / `logout` / `status`	Sign in (browser, `--no-browser` code, or `--token`), sign out, who am I (terminal only)
`rd config show`	Where the saved choices live (terminal only)

Pipelines

Command	What it does
`rd pipelines list`	Pipelines with schedule, status and last sync
`rd pipelines sync <name\|id>`	Start a run now
`rd pipelines pause <name\|id>` / `resume`	Stop or restart the scheduled runs

dbt

Command	What it does
`rd dbt run [models]`	Run dbt (also `build`, `test`, `seed`, `compile`, `snapshot`)
`rd runs list [--limit n]`	Recent runs
`rd runs get <run>`	One run’s details
`rd runs logs <run> [--follow]`	A run’s log; `--follow` watches it live
`rd runs cancel <run>`	Stop a running run
`rd schedules list`	The project’s schedules
`rd schedules run <name\|id>` / `pause` / `resume`	Run a schedule now, or pause and resume it

dbt options: --exclude <models>, --full-refresh, --fail-fast, --committed (run the committed code instead of the working copy, as scheduled runs do) and --follow (wait and show the log, exit code 1 if it fails). Other dbt options aren’t allowed. A run is refused while another is in progress in the project.

Knowledge

Command	What it does
`rd knowledge list`	Documents, and which are available to the Agent
`rd knowledge get <title\|id>`	Print a document’s Markdown
`rd knowledge put <title> [--file doc.md] [--no-agent]`	Create or update (by title) from a file or from standard input
`rd knowledge rm <title\|id> --yes`	Delete (asks for `--yes`)

Activity and the Agent

Command	What it does
`rd activity [--since 1d\|7d\|30d\|all] [--app x] [--status x] [-q text] [--limit n]`	What ran in the project. Admins see everyone’s, editors their own
`rd ask "<question>" [--chat <id>]`	Ask the Agent; shows its steps, then the answer. Continue with `--chat`

SQL (SELECT only)

rd sql "select * from \`project.dataset.table\` where date >= '2026-01-01' limit 20"
rd sql --file report.sql --format csv > report.csv
echo "select 1" | rd sql --format vertical

Options: --rows <n> (default 100, at most 10,000), --format table|json|csv|vertical (or --json).

The rules:

One SELECT statement. BigQuery itself checks what the statement is, so anything else is refused: inserts, updates, deletes, creating or dropping tables, scripts, exports.
It runs as the project’s BigQuery service account, like the SQL runner in Explore, so it can read whatever that account can.
The project’s query size limit applies, and you’re told when a query would be over it.
Results are formatted as a table that fits your terminal; timestamps are shown in UTC.

Environment variables

Variable	Meaning
`RD_TOKEN`	An access token (instead of `rd auth login`)
`RD_PROJECT`, `RD_ACCOUNT`	Default project and account
`RD_API_URL`	The server (default `https://api.reflectivedata.com`)
`NO_COLOR`	Plain output without colours (`--no-color` does the same)

Sign-in and choices are saved in ~/.config/rd (or $XDG_CONFIG_HOME/rd, %APPDATA%\rd); the sign-in file is readable only by you.

Exit codes

0 success, 1 an error or a failed --follow run, 2 not signed in or the token is no longer valid.