datus-ai Statsig
datus statsig drives the Statsig Console API (version 20240601). Global
usage:
datus statsig [--profile <env>] <group> <subcommand> [args...]
--profile (before the group) selects the configured environment. Output is
JSON by default; add --compact for single-line JSON, or -o table|plain|yaml
for other formats. Reads run freely; every mutating command confirms before
running.
Rate limits: mutations are capped at ~100 requests / 10s and ~900 / 15min per project — batch write-heavy work and expect HTTP 429 if you exceed them.
Metrics
datus statsig metrics list [-t TAG] [--show-hidden] [--limit N]
datus statsig metrics get <id> # or: --name NAME --type TYPE
datus statsig metrics sql <id> # the SQL Statsig generates for a metric
datus statsig metrics values --date 2024-09-01 [--metric-name N] [--metric-type T]
datus statsig metrics create --json '{...}' [--dry-run] # confirms
datus statsig metrics update <id> --json '{...}' [--dry-run] # confirms
datus statsig metrics reload <id> [--incremental] # WHN recompute, confirms
metrics sql is the fastest way to see the warehouse SQL behind a metric.
Warehouse-native metric sources (SQL authoring)
datus statsig metric-source list [--limit N]
datus statsig metric-source get <name>
datus statsig metric-source create --json '{...}' [--dry-run] # confirms
datus statsig metric-source update <name> --json '{...}' [--dry-run]
datus statsig metric-source delete <name> # confirms
Composing a write body
Write commands take the request body as --json '<inline>' or
--json-file <path> (bash cannot pipe stdin). To learn the body shape without
triggering a confirmation prompt, run describe (read-only):
datus statsig describe metric-source create
datus statsig describe metrics create
describe prints the same annotated template shown in each write command's
--help. Prefer --dry-run (adds "dryRun": true) to validate a body before
persisting. Missing required fields fail fast with exit 2 naming the field.
Example:
datus statsig metric-source create --json '{"name":"purchases","sql":"SELECT user_id, ts, amount FROM prod.orders","timestampColumn":"ts","idTypeMapping":{"userID":"user_id"}}' --dry-run
Experiments (results / analysis)
datus statsig experiments list [--status S] [-t TAG] [--limit N]
datus statsig experiments get <id>
datus statsig experiments pulse <id> --control <group> --test <group> [--cuped C] [--confidence X] [--date D]
datus statsig experiments summary <id> [--control G] [--test G]
datus statsig experiments exposures <id> [--dimensional] [--dimension-type T] [--severity S]
datus statsig experiments load-pulse <id> [--refresh] # trigger WHN Pulse compute, confirms
datus statsig experiments pulse-status <id> [<dag_id>] # pulse load history / one job
Typical WHN flow: experiments load-pulse <id> → poll
experiments pulse-status <id> → read experiments pulse <id> --control ... --test ....
ETL ingestion
datus statsig ingestion runs [--limit N]
datus statsig ingestion run <id>
datus statsig ingestion status --start 2024-09-01 --end 2024-09-08 [--source S] [--dataset D] [--status ST]
datus statsig ingestion get --type metrics --dataset <dataset> [--source-name N]
datus statsig ingestion schedule-get --dataset <dataset>
datus statsig ingestion backfill --type metrics --dataset <dataset> --start 2024-09-01 --end 2024-09-07 [--source S] # confirms
datus statsig ingestion schedule-set --dataset <dataset> --hour 3 # scheduled_hour_pst, confirms
Warehouse connections, events, logs, reports
datus statsig warehouse-connections update --json-file conn.json # credentials; file only, confirms
datus statsig events list [--limit N]
datus statsig events get <event_name> [--limit N]
datus statsig logs query [--query EXPR] [--source logs|events|spans] [--start MS] [--end MS] [--limit N] [--after CURSOR]
datus statsig reports get --type pulse_daily|first_exposures|topline_impact_daily --date 2024-09-01
warehouse-connections update takes credentials — always via --json-file
(never inline on the command line); delete the file afterwards.
Exit codes
0 success · 1 runtime/API error (incl. HTTP 429 rate limit) · 2 usage error (incl. missing required body field) · 3 profile/config error (e.g. no api_key) · 8 missing optional dependency.