The Omni CLI Is Here: Automate Omni Analytics from the Terminal

For most of Omni's life, automating the platform meant writing your own scripts against the REST API. That era is over: there's now an official Omni CLI — exploreomni/cli — that puts the entire API surface in your terminal, your CI pipelines, and your AI agents' hands.

Credit where it's due: the community got there first. Hawkfry Group — the Omni platform specialists we partner with — built an automation-first omni-cli with JSON-first output, stable exit codes, and machine-readable command discovery, explicitly designed "for agents, workflows, and machine-to-machine use." That repo now points at the official CLI as its successor, and the official tool carries the same automation-first DNA. If you want to understand where the ecosystem is heading, watching what Hawkfry builds is a good habit — their omni-resources hub is the best community collection of Omni guides and cheatsheets.

---

What the official CLI is

The design is unusually clever: commands are auto-generated from Omni's OpenAPI spec at build time. The spec is embedded in the binary; at startup, every API tag becomes a command group, path parameters become positional arguments, query parameters become flags. When Omni ships a new API endpoint, the CLI gains the command with a spec update and rebuild — no hand-written wrappers, no lag between API and tooling.

Practical consequences:

• Full API coverage by construction. Models, documents, users, and everything else the API exposes.
• JSON everywhere. All output is JSON to stdout, errors are JSON to stderr, and --compact makes it pipe-friendly for jq.
• Multiple profiles. omni config init sets up org, endpoint, and token; keep separate profiles for different orgs or environments.
• Two token types. Organization-wide tokens for admin automation, personal access tokens (PATs) for user-scoped work. Auth precedence is --token flag → OMNI_API_TOKEN env var → profile config.
• Shell completions for bash, zsh, fish, and PowerShell — including gems you might not know exist, like omni ai search-omni-docs.

Installing it

# Homebrew (macOS / Linux)
brew tap exploreomni/tap
brew install omni

# then
omni config init
omni models list
omni documents list

Pre-built binaries for macOS, Linux, and Windows are on the releases page, and there's an install script that verifies checksums if you prefer.

What to automate first

Where we see the CLI paying off immediately in real Omni deployments:

1. Admin operations in scripts, not clicks. User and group management is the classic case — anything you currently do one-at-a-time in the admin UI becomes a scriptable loop. For heavier lifting (bulk provisioning, IdP sync, user attributes), Hawkfry's omni-user-manager remains the purpose-built tool — we covered it in our Omni user management guide.

2. Content and model inventory. omni models list and omni documents list piped to jq gives you an instant audit surface: what exists, who owns it, what's stale. This is the foundation for the content-governance workflows we run on Looker instances, now reproducible on Omni.

3. CI checks on model changes. Because output is JSON with clean error semantics, wiring the CLI into a pull-request pipeline is straightforward — validate against the live instance before merging model changes, the same discipline we preach for dbt and LookML.

4. AI-agent access. This is the quiet headline. A JSON-first CLI with predictable errors and auto-discovery is exactly what an AI agent needs to operate a BI platform safely. Hawkfry's original CLI said this out loud — built "for agents, workflows, and machine-to-machine use" — and it's why we can drive Omni administration from Claude Code today. If you're exploring AI-operated analytics, this is the interface layer that makes it real.

The Looker comparison

Looker never got this. Its API is mature, but official tooling stayed at SDK level — every team ends up writing and maintaining its own admin scripts (we've written plenty). Omni shipping an official, spec-generated CLI this early says a lot about how the platform thinks about operability, and it compounds the argument we made in Omni vs Looker: the velocity difference is structural.

The ecosystem around it

The CLI is one piece of a fast-maturing toolchain, much of it community-driven:

• omni-resources — community-curated guides, cheatsheets, and best practices (Hawkfry Group)
• omni-user-manager — bulk user, group, and attribute sync (Hawkfry Group)
• omni-embed-demo — a TypeScript reference app for Omni's embedding SDK with signed-URL auth done right (Hawkfry Group)
• Looker-to-Omni migrator — our conversion engine for teams leaving LookML

If you're evaluating Omni or mid-migration, start with our resource guide for the full map.

---

Labs4Change partners with Hawkfry Group on Omni Analytics implementations — they run the platform side, we run migrations and the modeling layer. Book a free strategy call to talk through your Omni adoption or automation roadmap.