tenant

name

tenant

description

Look up ONE Confluence tenant (by domain / slug / cloudId) and answer "is this a paying customer, how big are they, what state are they in" — across all ZenUML apps (Full / Lite / Diagramly), including the Lite Stripe/KV space-license layer. Use whenever the question is about a SINGLE tenant: "is <domain> a paying customer", "look up / who is <domain>", "tenant profile", "how big is <domain>", "is X on a trial / when does the trial expire", or a tenant's paid status for a paywall / conversion / extension-request decision. For cohort / rollup / portfolio questions (total revenue, renewals due in a window, overdue payers, top payers) use the `marketplace-audit` skill instead. Discriminator: one domain in the question -> this skill; a cohort/time-window/ranking -> marketplace-audit. Triggers on tenant, whois, "is <domain> paying", paid status, tenant lookup, cloudId, tenant profile, tenant size, trial expiry.

Tenant lookup

Answers per-tenant questions off Atlassian Marketplace license/transaction data plus the Lite Stripe/KV space-license layer. It's the single-tenant primitive over marketplace-audit's engine (mp_report.py) — same auth, cloudId join, and SQLite snapshot — so there is no separate code to drift. (The script physically lives under marketplace-audit/scripts/; this skill references it by path, the same way macro-count references mixpanel's script.)

The one command you want

S=.claude/skills/marketplace-audit/scripts/mp_report.py
python3 $S whois <domain>          # the card: cloudId, users, per-app STATE, lifetime $, + Lite Layer-B KV
python3 $S whois <domain> --json   # machine-readable (self-documents data vintage in --local mode)

Always read whois before concluding anything about a tenant's paid status — a naive Marketplace $0 is the single most common wrong answer here.

What the card encodes (the traps baked in)

• State per app, not a binary. FREE (Lite free rider) / TRIAL — expires D (⚠ Nd left) (active evaluation) / PAID $X paid_thru D / LAPSED (was paying, coverage ended). A $0 is NOT "not a customer": it may be an active trial (a hot conversion, e.g. a Full eval expiring in 2 days) or a churned payer.
• Lite → Layer B auto-checked. Lite is free on the Marketplace ($0 by design); real Lite paid status lives in the Stripe/KV space-license layer. whois shells wrangler kv … --remote (the --remote is baked in — without it wrangler v4 reads LOCAL state and returns a false empty) and prints [Layer B: none / N space-licenses]. --no-kv skips it (~2s).
• cloudId is the join key. Transactions carry cloudId + company but NO hostname, so a text search by hostname finds the license yet misses every transaction → a false $0. whois joins on cloudId; do not hand-roll a text search.
• Slug typo → suggestions. A miss prints near-match hosts (woolworth-agile → woolworths-agile).

Drill-downs (same script)

python3 $S client <name>   # full license + transaction history for one tenant ("did they ever pay, how much, what plan")
python3 $S tier <domain>   # quick tier / license-type / status

Determinism / offline (optional)

python3 $S sync builds a local SQLite snapshot; then add --local to any command for byte-deterministic, offline, sub-100ms lookups. Billing may be stale (it prints the snapshot age and warns past 24h); Lite Layer-B is skipped in --local mode and says so. Use --local for batch / reproducible reports, live for "is X paying right now". See marketplace-audit for sync details.

Related

• marketplace-audit — the cohort/rollup side: revenue, renewals due, overdue/lapsing, top payers.
• paywall / extend-space-license — act on a Lite tenant's paid status once you know it.
• metrics / macro-count — per-space KV / usage: whether the tenant is using the app (vs paying).