M
MCP Directory CURATED DIRECTORY
911 MCP servers indexed
中文
← Back to directory
G

Gemini CLI

Community
Not a standard server
An open-source AI agent for Gemini in your terminal

This is not a typical connect-and-use MCP server; read the source repository before using it.

How to install ↓ Compare GitHub source repository ↗
Category
Dev Tools #147 of 438
Stars
★ 107.4k Very popular
Runtime
Node.js
Credentials
Optional API key
License
Apache-2.0
Last commit
58FMRS · C

Based on the supplied README, Gemini CLI is an Apache-2.0 open-source terminal AI agent with MCP client extensibility. The source does not provide evidence for a standalone MCP server manifest, transport, tool list, or client compatibility.

Strongest · Maintenance 16/20 Weakest · Reliability 8/20

Reliability
8/20
Security and permissions
8/20
Maintenance
16/20
Documentation
14/20
Setup experience
12/20
Why each score
Reliability 8/20
Only the README is provided; there is no server manifest, source code, MCP initialization/tool-list handshake evidence, or tests covering key paths. The README says Gemini CLI supports MCP, but the actual startup behavior and declared tool behavior for the repository root as an MCP server cannot be verified. The basic path is therefore plausible, but substantial points are deducted for missing reproducible execution and error-handling evidence.
Security and permissions 8/20
The README describes file operations, shell commands, web fetching, Google Search, and MCP extensions, indicating potentially broad permissions and external data boundaries. It links to sandbox, trusted-folder, telemetry, and authentication documentation, but those controls are not evidenced in the supplied material. Examples use placeholders rather than real credentials, so no red-line issue is shown; points are deducted for unverified least privilege, confirmation mechanisms, and data-flow disclosure.
Maintenance 16/20
The project identifies an Apache-2.0 license and provides stable, preview, and nightly release channels, links to CI, a roadmap, contribution guidance, issue tracking, and security advisories, with Google and the open-source community identified as maintainers. This supports clear ownership, update paths, and security-response channels. Full marks are not justified because the supplied material contains no actual commit history, release records, dependency updates, or issue-response evidence.
Documentation 14/20
The README covers installation, several authentication options, basic commands, headless operation, MCP configuration location, capabilities, release channels, troubleshooting, and security-related documentation, with multiple examples. Points are deducted because the supplied material lacks a manifest, concrete tool parameters and protocol details, a complete mainstream-client configuration, and reviewable cost and limitation details; several claims are delegated to external links.
Setup experience 12/20
The README offers npx, npm, Homebrew, MacPorts, and Anaconda installation paths, along with authentication and basic CLI usage examples, making initial setup reasonably clear for common developer environments. Points are deducted because there is no MCP server manifest, standard client connection configuration, runtime-version evidence, or handshake verification. The supplied material also contains no committed tests or CI evidence covering the server's key paths.

Static review · not runListed 2026-07-29

Read the FMRS scoring method →

Fit and risk

What it can accessReads local filesWrites / deletes local filesRuns commands or codeUses the network

Best for

  • Developers who work primarily in the command line
  • Individual developers and teams needing code understanding, generation, and automation
  • Users who want to connect custom MCP tools to Gemini CLI

Not for

  • Users looking for a documented MCP tool catalog
  • Users needing a standalone MCP server to deploy and connect directly
  • Environments that do not permit terminal, file, or shell capabilities

Required permissions

  • Gemini authentication through a Google account, Gemini API key, or Vertex AI configuration
  • File and shell access when the corresponding built-in tools are used
  • Network access for Google Search grounding and web fetching
  • GOOGLE_CLOUD_PROJECT may be required for paid organizational Code Assist licenses

Risks and side effects

  • File operations and shell commands can affect the project or runtime environment; sandboxing and trusted-folder controls should be reviewed
  • Preview and nightly channels may contain regressions or outstanding issues
  • API keys, Google accounts, and cloud project configuration require protection

Setup

Before you start

Runtime：Node.js

GEMINI_API_KEY optionalsecret Gemini API key for the API-key auth option, obtained from https://aistudio.google.com/apikey; alternatively use Google OAuth sign-in.
GOOGLE_API_KEY optionalsecret Google Cloud API key for the Vertex AI auth option, used together with GOOGLE_GENAI_USE_VERTEXAI=true.
Other optional settings (2)
GOOGLE_GENAI_USE_VERTEXAI optional Set to true to enable Vertex AI authentication, intended for enterprise/production workloads.
GOOGLE_CLOUD_PROJECT optional Google Cloud project ID required when using a paid Code Assist License.

Run npx @google/gemini-cli for an instant launch. Alternatively, install with npm install -g @google/gemini-cli, or use brew install gemini-cli on macOS/Linux. Start it with gemini. Authentication supports Google account sign-in, GEMINI_API_KEY, or Vertex AI environment configuration.

Check that it works

Run `npx @google/gemini-cli` or `gemini` in a terminal; if the interactive session starts and offers Google sign-in or API key authentication, the installation works.

Troubleshooting

  1. Verify Google sign-in or the relevant GEMINI_API_KEY, GOOGLE_API_KEY, and GOOGLE_GENAI_USE_VERTEXAI settings
  2. For a paid Code Assist license, verify GOOGLE_CLOUD_PROJECT is set
  3. Check the release channel; use latest when a stable release is required
  4. MCP server configuration is documented at ~/.gemini/settings.json
  5. Consult the troubleshooting guide or use the /bug command

Things to try

Once connected, you can ask your AI assistant things like:

  • Give me a summary of all of the changes that went in yesterday
  • Explain the architecture of this codebase
  • Write me a Discord bot that answers questions using a FAQ.md file I will provide
  • Run tests and deploy

Use cases

Analyzing and editing large codebases from the terminal
Debugging and troubleshooting with natural language
Automating tasks through scripts or GitHub workflows
Adding custom capabilities by configuring MCP servers

Overview

Gemini CLI is an open-source AI agent that brings Gemini directly into the terminal. It supports code understanding and generation, file operations, shell commands, web fetching, Google Search grounding, and MCP-based custom integrations. The repository documents a terminal AI agent with MCP client capabilities; the source does not establish that it exposes a standalone MCP server or a tool catalog.

Similar servers

Context7 80 · B

Upstash's official server providing up-to-date third-party library docs for AI coding assistants

★ 63.3k · Tools 2 Compare with this →
Compare with other MCP servers

Source revision ed96faedfd55 Data synced 2026-10-19 Read the FMRS scoring method