---
title: "Tiger CLI: One Command Line for Tiger Cloud"
description: "Tiger CLI gives you explicit, repeatable commands for your whole Tiger Cloud fleet. Tiger MCP ships in the same binary, ready for your AI agent."
url: "https://www.tigerdata.com/tiger-cli"
---


# All of Tiger Cloud, one command line

Tiger CLI gives you explicit, repeatable commands for your whole Tiger Cloud fleet: create, resize, connect, and script it in CI. Tiger MCP ships in the same binary and gives your AI agent its own way in through natural language

```bash
curl -fsSL https://cli.tigerdata.com | sh
```

Supported: mac os, linux, windows · brew, apt, yum

[Read the docs](https://www.tigerdata.com/docs/get-started/quickstart/tiger-cli)

FROM ZERO TO RUNNING POSTGRES IN 4 COMMANDS

```bash
$ tiger auth login
Authenticated. Session stored in system keychain.

$ tiger service create --name sensors --region us-east-1
Creating service 'sensors'… ready in 28s

$ tiger db connect sensors
psql (17.4) — connected

$ tiger mcp install claude-code
Tiger MCP configured for claude-code ✓
```

## From install to a tuned database

### onboarding — Create, then configure

A running service is step one. Set up hypertables, columnstore, and data tiering from the same tool.

### lifecycle — Full service management

Create, resize, start, stop, and delete services. VPCs, read replicas, and credentials included.

### automation — Explicit and repeatable

Every command does exactly what it says, every time. Commit it to a script or a CI/CD pipeline.

### auth — Secure by default

OAuth sign-in with sessions in your system keychain. Client credentials for scripts and CI.

### agents — MCP in the same binary

Tiger MCP install configures Claude Code, Cursor, and more. No separate install, ever.

### pricing — Free to try

tiger service create --cpu shared --memory shared launches a free service in beta.

## Every step runs from your terminal or your agent

Run each stage as a CLI command, or ask Tiger MCP in plain English. Same services, same permissions, same results either way.

01. **set up** — Install, authenticate, connect your agent.
02. **configure** — Add hypertables, columnstore, and tiering policies.
03. **validate** — Ask Tiger MCP to review a schema against best practices.
04. **operate** — Run queries, rotate credentials, resize.
05. **debug** — Pull logs and run diagnostic queries.

## Your AI agent, wired to your database

Where the CLI executes, Tiger MCP reasons. It gives an agent typed tools to manage services and run SQL, plus what no CLI command can do: design schemas, plan migrations, and recommend improvements, backed by built-in skills and search across Tiger Data documentation.

Hook it up to an agent and onboarding runs on autopilot: it designs the schema, creates the hypertables, enables columnstore, and sets tiering policies while you review.

You stay in control: read-only mode blocks the agent's writes, DDL, and infrastructure changes. Keeping your database safe.

```bash
$ tiger mcp install claude-code

Tiger MCP configured ✓
```

then, in your agent

```
"Which of my tables are hypertable candidates, and what should the chunk interval be?"

"Set up columnstore and data tiering on sensor_readings."
```

Works with: CLAUDE-CODE, CURSOR, CODEX, GEMINI-CLI, VS-CODE

## Install once. Run everything.

One open-source binary for you and your agent.
Free Tiger Cloud services available to start.

```bash
curl -fsSL https://cli.tigerdata.com | sh
```

[Get started guide](https://www.tigerdata.com/docs/get-started/quickstart/tiger-cli) · [Star on Github](https://github.com/timescale/tiger-cli)