pkg.sopackage field notes

brew / approval gates / rank 69

Install gemini-cli with Homebrew

Interact with Google Gemini AI models from the command-line. Version 0.46.0 via Homebrew; verified 2026-06-15.

agent safety

Agent safety answer

gemini-cli is an AI CLI that can combine model access with local files and tool calls.

Credential access

Reads model API credentials, environment variables, and files included in prompts.

Remote mutation

Can drive local commands or API calls if granted tool access.

Publish/artifact risk

Can generate code and release commands that become artifacts.

Recommended control

Run in containment and gate all secret-backed or mutating commands.

Agent-use guidance

Limit file scope and require approval before exposing secrets or running deployment tools.

install

Additional install commands

macOS

Homebrewverified · 100%
brew install gemini-cli

provider-native install command

overview

Package summary

Interact with Google Gemini AI models from the command-line

Commands and aliases

  • gemini

history

Project history and usage

Gemini CLI is Google's open-source terminal agent for interacting with Gemini models from the command line. It became significant quickly because Google launched it as an Apache-2.0 project with free individual access, Code Assist integration, built-in tools, and package-manager distribution through npm and Homebrew.

Project history

Google announced Gemini CLI on June 25, 2025 as an open-source AI agent that brings Gemini into developers' terminals. The launch positioned the CLI as a local utility for coding, content generation, research, problem solving, and task management rather than only as a chat wrapper.

The public repository developed around a terminal-first workflow: installation through npm or Homebrew, documented release channels, configuration through settings files and `GEMINI.md`, support for non-interactive invocation, and integration points such as MCP servers and extensions.

Adoption history

Gemini CLI's adoption path was tied to Google's Gemini Code Assist product line. The launch article stated that the tool shared technology with Gemini Code Assist, giving developers prompt-driven AI help in both VS Code and the terminal.

Package-manager adoption followed the developer-tool pattern: the README documents global npm installation, `npx` execution, Homebrew installation, and MacPorts packaging. That made the CLI visible both to JavaScript users and to developers who expect AI tools to install like ordinary command-line utilities.

How it is used

Practitioners use Gemini CLI inside source trees to ask questions about code, edit files, run shell commands, debug failures, generate applications from multimodal inputs, and automate tasks from scripts. The tool also supports Google Search grounding, file operations, web fetching, conversation checkpointing, and project-specific context files.

Teams use the CLI alongside Gemini Code Assist and GitHub workflows, while individual users often treat it as a terminal-native coding assistant that can operate without leaving the shell.

Why package nerds care

For package nerds, Gemini CLI is a marker of AI coding tools becoming ordinary CLI packages. It is distributed through language-package tooling and system package managers, has documented stable, preview, and nightly release channels, and exposes configuration paths that matter for reproducible developer environments.

Timeline

  • 2025-06-25: Google announces Gemini CLI as an open-source AI agent for the terminal.
  • 2025: The repository publishes early preview, stable, and nightly release-channel tags.
  • 2025: Homebrew, MacPorts, npm, and other package-manager entries make the tool installable through common developer workflows.

Related projects

  • Gemini CLI is related to Gemini Code Assist, Google AI Studio, Vertex AI, MCP servers, and the broader family of terminal AI agents used for code navigation and automation.

approval gates

Human review metadata for risky commands

The local approval-gate seed includes 5 rules for gemini-cli. Covered entrypoints: gemini. Severity labels: critical, high, medium. Coverage: partial, reviewed 2026-05-21.

Example gated actions

  • Send a prompt to a remote Gemini model.
  • Include broad workspace file context in a Gemini session.
  • Run with reduced approval prompts or automatic tool execution.
  • Enable checkpointing that writes session state to disk.
  • Configure or use MCP servers from the Gemini CLI.

local files

Configuration and credential file locations

These source-backed paths show where this package keeps local settings or durable credentials. Automic Vault can use them as review targets for secret scanning, migration, and command approval.

Configuration files

Config paths the tool may read or write during local use.

Linux
~/.gemini/settings.json.gemini/settings.json/etc/gemini-cli/settings.json/etc/gemini-cli/system-defaults.json
macOS
~/.gemini/settings.json.gemini/settings.json/Library/Application Support/GeminiCli/settings.json/Library/Application Support/GeminiCli/system-defaults.json
Windows
~/.gemini/settings.json.gemini/settings.jsonC:\ProgramData\gemini-cli\settings.jsonC:\ProgramData\gemini-cli\system-defaults.json

executables

Installed executables

CommandKindExposureNote
geminiexecutableindexed executableDiscovered from the local executable index.

freshness

Version and freshness

These signals separate page generation age, package-manager activity, and upstream release comparison. Version lag is warned only when an evidence URL and comparable versions are present.

page generated2026-08-03
manager version0.46.0
manager updated2026-06-15
local dataunknown
upstreamnot available
latest detectednot detected
  • okNo freshness warnings were generated.

install metadata

Package metadata

Package keybrew:gemini-cli
Version0.46.0
Package managerHomebrew
Homepagehttps://geminicli.com
Last updated2026-06-15T10:20:17-04:00
Pulseupdated
Bottlenot recorded
Servicenone declared

source trail

Generated from repository data

This page is generated by av-web from the private package SQLite artifact built by scripts/generate-pkg-sqlite.py.

Used sources

  • Geiger risk classifier
  • Nucleus package database
  • approval-gate seed metadata
  • curated agent safety answer
  • curated configuration and credential file locations
  • curated package history
  • pkgdb category and tag curation