getop 1.0.1

CLI for Google Gemini Enterprise

getop

getop is a command-line tool for Google Gemini Enterprise administrators.

getop allows you to:

• Inventory everything deployed — engines, data stores, connectors, agents, and license seats in one overview
• Monitor connector health — sync state, freshness, and errors across every collection
• Investigate prompts and replies (if logging is enabled) — per user or fleet-wide, live-tailable
• See what Model Armor caught — jailbreak, RAI, and malicious-URI hits, plus the active policy
• Stay ahead of quota — usage vs limits with one-click links to the Cloud console
• Run a health check — a parallel PASS/WARN/FAIL sweep with CI-friendly exit codes

Read-only by design — every command runs with viewer roles alone.

$ getop info
             Gemini Enterprise — acme-search-prod (global)

╭──── Engines ─────╮  ╭── Data stores ───╮  ╭──── Connectors ────╮
│        2         │  │        9         │  │     2/3 ACTIVE     │
╰──────────────────╯  ╰──────────────────╯  ╰────────────────────╯
╭───── Agents ─────╮  ╭───── Licenses ─────╮
│        41        │  │   418/500 (84%)    │
╰──────────────────╯  ╰────────────────────╯

╭────────────── Support Search ──────────────╮  ╭───────────────── Sandbox ──────────────────╮
│ SEARCH · GENERIC · intranet · 2026-04-08   │  │ SEARCH · GENERIC · intranet · 2026-03-26   │
│ Data stores (4)                            │  │ Data stores (3)                            │
│   • sharepoint_1774_file  ← sharepoint     │  │   • sharepoint_1774_file  ← sharepoint     │
│   • sharepoint_1774_page  ← sharepoint     │  │   • onedrive_1775_file    ← onedrive       │
│   • onedrive_1775_file    ← onedrive       │  │   • gdrive_1778_file      ← google_drive   │
│   • jira_1778_issue  ← jira  PAUSED        │  │ Agents (4 — 4 enabled, 0 defaults)         │
│ Agents (41 — 4 enabled, 37 defaults)       │  │   • Security Reviewer   ENABLED            │
│   • Deep Research      ENABLED             │  │   • Deep Research       ENABLED            │
│   • Contract Analyst   ENABLED             │  │   • Idea Generation     ENABLED            │
│   • My Agent ×37 (user defaults)           │  │ Features (14 on · 7 off)                   │
│ Features (16 on · 5 off)                   │  │   ✓ agent-gallery model-selector           │
│   ✓ agent-gallery model-selector           │  │   ✗ session-sharing onedrive-upload        │
│   ✗ agent-sharing-without-approval         │  ╰──────────── sandbox_1774543712 ────────────╯
╰──────── support-search_1775663018 ─────────╯

Two environments side by side. At a glance: prod's Jira connector is PAUSED (the 2/3 tile), licenses are at 84%, and sandbox has session-sharing and onedrive-upload disabled where prod allows them.

By design, the current release is strictly read-only — every command works with viewer roles alone, so it can be handed to anyone on the team without change-risk. It may grow administrative verbs (e.g. triggering connector syncs, managing agents) in a future release.

Contributing with an AI coding agent? Read AGENTS.md — it maps the project subagents, skills and hard constraints.

Install

pipx install getop

Also works with uv tool install getop or pip install getop. See Authentication for the one-time credential setup.

Commands

getop info — project overview

A whole-project dashboard: summary tiles (engines, data stores, connector health, agents, and license seats / activation / unmet demand) plus a card per engine showing its data stores with their connector sources, its agents ("My Agent" user defaults collapsed into a single ×N line), and its feature toggles. Diffing two environments' cards is the fastest way to spot config drift. Output is shown at the top of this page.

getop ls — inventory

getop ls engines|datastores|connectors|agents|licenses

Lists resources across the collection hierarchy: engines and data stores under default_collection, data connectors across all collections (each connector-backed source lives in its own), agents per engine, and user licenses.

$ getop ls connectors
                               Data Connectors
┏━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━┳━━━━━━━━┳━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Collection           ┃ Data Source ┃ State  ┃ Refresh Interval ┃ Entities ┃
┡━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━╇━━━━━━━━╇━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ acme-sharepoint_1774 │ sharepoint  │ ACTIVE │ 86400s           │ 5        │
│ acme-onedrive_1775   │ onedrive    │ ACTIVE │ 86400s           │ 1        │
└──────────────────────┴─────────────┴────────┴──────────────────┴──────────┘

$ getop ls licenses
                                 User Licenses
┏━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┳━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━┓
┃ User              ┃ State    ┃ Config      ┃ Last Login          ┃
┡━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━╇━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━┩
│ [e-mail] │ ASSIGNED │ enterprise  │ 2026-07-03T12:47:25 │
│ [e-mail]   │ ASSIGNED │ enterprise  │ 2026-07-03T11:25:22 │
└───────────────────┴──────────┴─────────────┴─────────────────────┘

getop logs — view and tail logs

getop logs connector [--datastore ID] [--severity ERROR] [--since 1h]
getop logs user [email] [--since 24h] [--follow]
getop logs ai [--since 24h] [--follow]

connector shows sync activity. user shows end-user activity — prompts, replies, searches, and Model Armor events — for one user or all. ai streams the raw gen_ai content logs (no identity field; use user to scope by principal). --follow/-f tails any stream; an empty result says whether logging is off or just unmatched.

$ getop logs user --since 24h
╭──────────────────────────────── SENSITIVE ─────────────────────────────────╮
│ ⚠  Output may include end-user prompt/response content if prompt/response   │
│ logging is enabled on this project.                                         │
╰─────────────────────────────────────────────────────────────────────────────╯
                          User activity: all users (24h)
┏━━━━━━━━━━┳━━━━━━━━━━┳━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━┓
┃ Time     ┃ Severity ┃ User            ┃ Message               ┃ Reply (truncated)   ┃
┡━━━━━━━━━━╇━━━━━━━━━━╇━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━┩
│ 16:57:06 │ INFO     │ alice@example.… │ StreamAssist: What's  │ Your travel policy  │
│          │          │                 │ our travel policy?    │ allows economy fa…  │
│ 12:49:04 │ WARNING  │ [e-mail] │ ModelArmorAudit:      │                     │
│          │          │                 │ Ignore all previous…  │                     │
└──────────┴──────────┴─────────────────┴───────────────────────┴─────────────────────┘

The WARNING row is Model Armor flagging a prompt-injection attempt. Tail it live with getop logs user -f, or pull full transcripts with --json.

getop armor — Model Armor violations

getop armor [--since 24h] [--all]

Surfaces prompts and responses that Model Armor flagged, with the filters that tripped and their confidence (jailbreak, RAI categories, CSAM, malicious URIs). Violations only by default; --all includes clean screenings. Carries no user identity — pair with getop logs user.

Add --summary for a per-filter rollup — hit counts, last seen, and an example input per category — instead of the event-by-event list:

$ getop armor --summary --since 7d
              Model Armor hits by filter (7d)
┏━━━━━━━━━━━━━━━━━━┳━━━━━━┳━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━┓
┃ Filter           ┃ Hits ┃ Last seen           ┃ Example input       ┃
┡━━━━━━━━━━━━━━━━━━╇━━━━━━╇━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━┩
│ pi_and_jailbreak │ 3    │ 2026-07-03T12:49:03 │ Ignore all previou… │
│ malicious_uris   │ 2    │ 2026-07-03T09:14:00 │ http://testsafebro… │
└──────────────────┴──────┴─────────────────────┴─────────────────────┘

Add --policy to print the configured Model Armor template(s) instead — the filters that are enabled and at what confidence:

$ getop armor --policy
            Model Armor policy — ge-default-amer (us)
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━┓
┃ Filter                            ┃ Enforcement ┃ Confidence       ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━┩
│ Prompt injection & jailbreak      │ ENABLED     │ HIGH             │
│ Responsible AI: DANGEROUS         │ ENABLED     │ MEDIUM_AND_ABOVE │
│ Malicious URIs                    │ ENABLED     │                  │
└───────────────────────────────────┴─────────────┴──────────────────┘

$ getop armor --since 7d
                          Model Armor violations (7d)
┏━━━━━━━━━━┳━━━━━━━━━━━┳━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━┓
┃ Time     ┃ Direction ┃ Match       ┃ Filters                   ┃ Content        ┃
┡━━━━━━━━━━╇━━━━━━━━━━━╇━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━┩
│ 12:49:03 │ prompt    │ MATCH_FOUND │ pi_and_jailbreak(HIGH),   │ Ignore all     │
│          │           │             │ rai:dangerous(HIGH)       │ previous inst… │
└──────────┴───────────┴─────────────┴───────────────────────────┴────────────────┘

getop stats — metrics

getop stats [--engine ID] [--since 24h]

Discovers the project's discoveryengine.googleapis.com metrics at runtime and summarises query volume, latency and connector sync freshness over the window.

$ getop stats --since 24h
                 Discovery Engine metrics — acme-search-prod (global)
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┳━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━┓
┃ Metric type                       ┃ Category  ┃ Points ┃ Aggregate           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━╇━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━┩
│ …/serving/search_request_count    │ volume    │ 288    │ 1,204 (sum)         │
│ …/serving/search_latencies        │ latency   │ 288    │ 342 (p95)           │
│ …/dataconnector/synced_doc_count  │ connector │ 24     │ 48,102 (sum)        │
└───────────────────────────────────┴───────────┴────────┴─────────────────────┘

getop quota — quota usage

getop quota [--since 24h]

Pairs each Discovery Engine quota's latest usage with its limit per location: percent used (highlighted at ≥75% / ≥90%), byte quotas in human units, and counts of quota-exceeded events over the window — the quickest way to spot the next capacity ceiling before ingestion hits it. On terminals that support hyperlinks, each quota name links to the project's Cloud console quotas page.

$ getop quota
                Discovery Engine quotas — acme-search-prod (global)
┏━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┳━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━┳━━━━━━━━━━┓
┃ Quota                  ┃ Location ┃ Usage     ┃ Limit      ┃ Used ┃ Exceeded ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━╇━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━╇━━━━━━━━━━┩
│ document_size_tier_st… │ global   │ 85.6 MiB  │ 1.0 GiB    │ 8.4% │          │
│ data_stores_regional   │ global   │ 8         │ 100        │ 8.0% │          │
│ documents_regional     │ global   │ 76        │ 10,000,000 │ 0.0% │          │
└────────────────────────┴──────────┴───────────┴────────────┴──────┴──────────┘

getop doctor — health check

getop doctor [--since 24h]

Runs the whole suite concurrently and renders a live PASS/WARN/FAIL table across inventory reachability, connector states and sync freshness, connector/API error logs, and metric availability. Exits non-zero if any check fails, so it drops straight into CI or cron.

$ getop doctor
                     getop doctor — acme-search-prod (global)
┏━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓
┃ Check                  ┃ Status ┃ Detail                                    ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┩
│ engines                │ OK     │ 1 engine(s): support-search               │
│ datastores             │ OK     │ 6 data store(s)                           │
│ data connectors        │ OK     │ 2 connector(s), all ACTIVE, synced <24h   │
│ agents                 │ OK     │ 37 agent(s) across 1 engine(s)            │
│ connector errors (24h) │ OK     │ no connector ERROR logs since 24h         │
│ API errors (24h)       │ OK     │ no discoveryengine API ERROR logs         │
│ monitoring metrics     │ OK     │ 369 metric type(s) discovered             │
└────────────────────────┴────────┴───────────────────────────────────────────┘

getop version

$ getop version
getop 0.2.2 (v0.2.2, commit a1854e8)

Prints the release, tag and the git commit the package was built from.

getop update

$ getop update
Update available: 0.2.2 → 0.3.0
$ pipx upgrade getop

Checks PyPI for a newer release and upgrades in place, detecting whether you installed via pipx, uv or pip. Use --check to only report.

Global options

Available on every command:

Switch	Meaning
--project / -p	Target GCP project. Defaults to the ADC project; also reads GOOGLE_CLOUD_PROJECT.
--location / -l	Gemini Enterprise location (default global). Regional values route to {location}-discoveryengine.googleapis.com.
--quota-project	Project to bill API quota against when you lack serviceusage.services.use on the target; also reads GOOGLE_CLOUD_QUOTA_PROJECT.
--json	Machine-readable output on every command (newline-delimited JSON in --follow mode).
--since	Look-back window on time-based commands: 30m, 1h, 24h, 7d.

Every table above is also available as JSON — pipe it straight into jq:

getop quota --json | jq '.[] | select(.percent_used > 75)'
getop ls licenses --json | jq '[.[] | select(.last_login_time == null)] | length'

Authentication

Authenticate once with Application Default Credentials:

gcloud auth application-default login

getop never reads or writes key files. Each command needs only a viewer role:

Role	Used by
roles/discoveryengine.viewer	getop ls …, getop info, getop doctor
roles/logging.viewer	getop logs …, getop doctor
roles/monitoring.viewer	getop stats, getop quota, getop doctor
roles/modelarmor.viewer	getop armor --policy

User credentials (as opposed to service accounts) also need a quota project: getop uses the target project automatically, which requires serviceusage.services.use there. If you don't have it, use --quota-project.

Enabling connector/observability logging on a project is a one-time setup step requiring roles/discoveryengine.agentspaceAdmin; getop only ever reads what's there.

License

MIT © Egen — see LICENSE.