mcpField
G
Gemini MCP Server

Gemini MCP Server

Local
github.com/chew-zAI & Agents2Updated 11w ago

About Gemini MCP Server

GeminiMCP is an MCP server that exposes configured DeepSeek or Qwen models for coding and analysis questions. It combines GitHub context, reusable prompts, server-side prompt selection, and HTTP or stdio transport.

Key features

  1. 01Answer coding and analysis questions
  2. 02Use GitHub PR, commit, diff, and file context
  3. 03Provide workflow and coding prompts
  4. 04Select system prompts by category
  5. 05Support HTTP and stdio transports

Use cases

  1. 01Ask coding questions with GitHub context
  2. 02Review a pull request
  3. 03Explain a commit or compare references
  4. 04Get help debugging, refactoring, or testing code

MCP support

Exposes a question-answering tool and workflow and coding prompts to MCP clients. Clients connect over JWT-secured HTTP or local stdio.

Configuration

Local server config is the same for every client. Copy it into your client’s MCP config file.

No install method listed yet

We haven’t collected a package or remote endpoint for this server. Follow the instructions in the project README.

Repository

Tools

No tool list yet

Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.

Details

From the project README

GeminiMCP

An MCP server that exposes configured DeepSeek or Qwen models to MCP clients (Claude Code, IDE extensions, custom tooling).

What it does

  • gemini_ask — coding/analysis question answering with composable GitHub context (PRs, commits, diffs, files)
  • 3 workflow prompts — review_pr, explain_commit, compare_refs
  • 7 coding prompts — code review, explain, debug, refactor, architecture, tests, security
  • Structured XML envelope — every user turn is rendered as <context> / <task> / <final_instruction> with attacker-controllable bodies emitted as raw text
  • Server-side system-prompt selection — a classifier picks a category-specific instruction and matching <final_instruction> body
  • Two transports: HTTP (JWT-secured, preferred) and stdio (local fallback)

Quick start

# Build
go build -o bin/mcp-gemini .

# Configure
cp .env.example .env
# Edit .env — set PROVIDER, PROVIDER_API_KEY, and PROVIDER_MODEL

# Run (stdio, for Claude Code)
./bin/mcp-gemini

# Run (HTTP)
./bin/mcp-gemini --transport=http

Claude Code mcpServers entry (stdio):

{
  "gemini": {
    "command": "/path/to/bin/mcp-gemini",
    "env": { "PROVIDER": "deepseek", "PROVIDER_API_KEY": "your-key", "PROVIDER_MODEL": "deepseek-v4-pro" }
  }
}

Documentation

DocContents
docs/QUICKREF.mdOne-page cheat sheet — parameters, env vars, CLI flags, examples
docs/TOOLS.mdFull tool and prompt reference with all parameters and examples
docs/PROMPTS.mdMCP-prompt registry and server-side system-prompt selection
docs/USAGE.mdInstallation, transport modes, authentication, all env vars
docs/ARCHITECTURE OVERVIEW.mdCodebase map, component diagram, request lifecycle
.env.exampleAnnotated configuration template

Development

./run_test.sh     # tests
./run_lint.sh     # golangci-lint
./run_format.sh   # goimports + gofmt
task check        # all gates via go-task (fmt+vet+lint+test)

Requirements

  • Go 1.26+
  • PROVIDER, PROVIDER_API_KEY, and PROVIDER_MODEL (DeepSeek or Qwen)
  • GEMINI_GITHUB_TOKEN (GitHub PAT — only for github_* parameters)

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.