Gemini Search MCP
# @kidapu/gemini-search-mcp
[](https://www.npmjs.com/package/@kidapu/gemini-search-mcp)
Gemini 3.1 Flash-Lite + Google Search Grounding を使った MCP サーバー。
**npm**: https://www.npmjs.com/package/@kidapu/gemini-search-mcp
## インストール
API キーは [Google AI Studio](https://aistudio.google.com/apikey) で取得できます。
### MCP サーバーとして追加(推奨)
```bash
# ユーザーレベル(全プロジェクトで使用可能)
claude mcp add gemini-search -s user -e GEMINI_SEARCH_MCP_API_KEY=your_api_key -- npx -y @kidapu/gemini-search-mcp
# プロジェクトレベル(このプロジェクトのみ)
claude mcp add gemini-search -s project -e GEMINI_SEARCH_MCP_API_KEY=your_api_key -- npx -y @kidapu/gemini-search-mcp
```
### プラグインとしてインストール
```bash
# 1. マーケットプレイスを追加
/plugin marketplace add kidapu/Gemini-Search-MCP
# 2. プラグインをインストール
/plugin install gemini-search@gemini-search-mcp
```
> **注意**: プラグイン方式では事前に環境変数の設定が必要です。
>
> ```bash
> # ~/.zshrc に追記
> export GEMINI_SEARCH_MCP_API_KEY="your_api_key"
>
> # 反映
> source ~/.zshrc
> ```
## 環境変数
| 変数名 | デフォルト | 説明 |
|--------|-----------|------|
| `GEMINI_SEARCH_MCP_API_KEY` | - | Google AI API キー(必須) |
| `GEMINI_MODEL` | `flash` | モデル: `flash`, `flash-lite`, `pro`, `2.5-flash` |
| `GEMINI_THINKING_BUDGET` | `-1` | Thinking: -1=auto, 0=off, 1-24576 |
| `GEMINI_WEB_SEARCH` | `true` | Web検索: true/false |
## モデル
| エイリアス | フルネーム | 特徴 |
|-----------|-----------|------|
| `flash` | `gemini-3.1-flash-lite-preview` | 最速・最安・高性能(デフォルト) |
| `flash-lite` | `gemini-3.1-flash-lite-preview` | ↑ と同じ |
| `pro` | `gemini-2.5-pro` | 最高性能 |
| `2.5-flash` | `gemini-2.5-flash` | 旧デフォルト |
| `2.5-flash-lite` | `gemini-2.5-flash-lite` | 旧最安 |
フルネームでも指定可能(例: `GEMINI_MODEL=gemini-3-flash-preview`)。
## 料金
| モデル | Input (1M tokens) | Output (1M tokens) |
|--------|-------------------|-------------------|
| `gemini-3.1-flash-lite-preview` | $0.25 | $1.50 |
| `gemini-3-flash-preview` | $0.50 | $3.00 |
| `gemini-2.5-flash` | $0.30 | $2.50 |
| `gemini-2.5-flash-lite` | $0.10 | $0.40 |
| `gemini-2.5-pro` | $1.25 | $10.00 |
> 最新の料金は [Gemini API Pricing](https://ai.google.dev/gemini-api/docs/pricing) を参照してください。
TDQS
Scored across 1 tool
With only one tool, there is no possibility of ambiguity or overlap between tools. The tool has a clearly defined purpose focused on AI-powered search with web grounding.
A single tool inherently has perfect naming consistency, as there are no other tools to compare against. The name 'gemini-search' follows a clear and descriptive pattern.
A single tool is too few for a server named 'Gemini Search MCP', which suggests a broader search functionality scope. While the tool is powerful, the set feels thin and limited, lacking complementary operations like filtering, saving results, or managing search history.
The tool surface is severely incomplete for a search domain. It provides a general search function but lacks essential operations such as refining searches, handling pagination, accessing cached results, or integrating with other data sources, which are typical in search APIs.