これは、Google の Gemini API と統合して、コードを生成せずに分析的思考機能を提供する Model Context Protocol (MCP) の実装です。
Gemini Thinking Serverは、GoogleのGeminiモデルを活用し、シーケンシャル思考と問題解決能力を提供する専用のMCPサーバーです。以下の機能を提供します。
複雑な問題を段階的に分解する
修正の余地がある計画と設計
軌道修正が必要な分析
当初は全体の範囲が明確でない可能性がある問題
Related MCP server: gpal
ジェミニパワー思考：ジェミニの分析能力を活用して思慮深い応答を生成します
メタ解説：推論プロセスに関する洞察を提供する
信頼度レベル: ジェミニの分析に対する信頼度を示します
代替パス：問題に対する異なるアプローチを提案する
思考の分岐：異なる思考経路の探索を可能にする
修正機能: 過去の考えの修正をサポート
セッションの永続性: 分析セッションを保存して再開する
# Clone the repository
git clone <repository-url>
# Install dependencies
npm install
# Build the project
npm run buildサーバーを実行する前に、Gemini API キーを設定する必要があります。
export GEMINI_API_KEY=your_api_key_herenode dist/gemini-index.jsgeminithinkingツールは次のパラメータを受け入れます。
query （必須）: 分析する質問または問題
context （オプション）: 追加のコンテキスト情報
approach （オプション）：問題に対する提案されたアプローチ
previousThoughts (オプション): コンテキストの以前の考えの配列
thought （オプション）：現在の思考ステップ（空の場合は、Geminiによって生成されます）
nextThoughtNeeded (必須): 別の思考ステップが必要かどうか
thoughtNumber （必須）: 現在の思考番号
totalThoughts （必須）: 必要な推定総思考数
isRevision （オプション）: 以前の考えを修正するかどうか
revisesThought （オプション）：どの考えが再考されているか
branchFromThought (オプション): 分岐点の思考番号
branchId (オプション): ブランチ識別子
needsMoreThoughts （オプション）: さらに考えが必要な場合
このツールは、セッション管理コマンドもサポートしています。
sessionCommand : セッションを管理するためのコマンド ('save'、'load'、'getState')
sessionPath : セッションファイルを保存または読み込むパス（'save'および'load'コマンドに必要）
{
"sessionCommand": "save",
"sessionPath": "/path/to/save/session.json",
"query": "dummy",
"thoughtNumber": 1,
"totalThoughts": 1,
"nextThoughtNeeded": false
}{
"sessionCommand": "load",
"sessionPath": "/path/to/load/session.json",
"query": "dummy",
"thoughtNumber": 1,
"totalThoughts": 1,
"nextThoughtNeeded": false
}{
"sessionCommand": "getState",
"query": "dummy",
"thoughtNumber": 1,
"totalThoughts": 1,
"nextThoughtNeeded": false
}ツールの使用方法の例を次に示します。
{
"query": "How might we design a sustainable urban transportation system?",
"context": "The city has 500,000 residents and currently relies heavily on personal vehicles.",
"approach": "Consider environmental, economic, and social factors.",
"thoughtNumber": 1,
"totalThoughts": 5,
"nextThoughtNeeded": true
}サーバーは次のように応答します:
{
"thought": "The generated thought from Gemini",
"thoughtNumber": 1,
"totalThoughts": 5,
"nextThoughtNeeded": true,
"branches": [],
"thoughtHistoryLength": 1,
"metaComments": "Meta-commentary about the reasoning",
"confidenceLevel": 0.85,
"alternativePaths": ["Alternative approach 1", "Alternative approach 2"]
}さまざまなユースケースを示すために、いくつかのサンプル クライアントが提供されています。
sample-client.js : 基本的なクライアントの例
example-usage.js : 具体的な使用例
codebase-analysis-example.js : コードベース分析の例
session-example.js : セッションの永続性を示す例
advanced-filtering-example.js : 高度なセマンティックフィルタリングを示す例
セッション例を実行するには:
node dist/session-example.js高度なフィルタリングの例を実行するには:
node dist/advanced-filtering-example.jsマサチューセッツ工科大学
A detailed tool for dynamic and reflective problem-solving through Gemini AI. This tool helps analyze problems through a flexible thinking process powered by Google's Gemini model. Each thought can build on, question, or revise previous insights as understanding deepens.
When to use this tool:
Breaking down complex problems into steps
Planning and design with room for revision
Analysis that might need course correction
Problems where the full scope might not be clear initially
Problems that require a multi-step solution
Tasks that need to maintain context over multiple steps
Situations where irrelevant information needs to be filtered out
Key features:
Leverages Gemini AI for deep analytical thinking
Provides meta-commentary on the reasoning process
Indicates confidence levels for generated thoughts
Suggests alternative approaches when relevant
You can adjust total_thoughts up or down as you progress
You can question or revise previous thoughts
You can add more thoughts even after reaching what seemed like the end
You can express uncertainty and explore alternative approaches
Not every thought needs to build linearly - you can branch or backtrack
Session persistence: save and resume your analysis sessions
Parameters explained:
query: The question or problem to be analyzed
context: Additional context information (e.g., code snippets, background)
approach: Suggested approach to the problem (optional)
previousThoughts: Array of previous thoughts for context
thought: The current thinking step (if empty, will be generated by Gemini)
next_thought_needed: True if you need more thinking, even if at what seemed like the end
thought_number: Current number in sequence (can go beyond initial total if needed)
total_thoughts: Current estimate of thoughts needed (can be adjusted up/down)
is_revision: A boolean indicating if this thought revises previous thinking
revises_thought: If is_revision is true, which thought number is being reconsidered
branch_from_thought: If branching, which thought number is the branching point
branch_id: Identifier for the current branch (if any)
needs_more_thoughts: If reaching end but realizing more thoughts needed
metaComments: Meta-commentary from Gemini about its reasoning process
confidenceLevel: Gemini's confidence in the generated thought (0-1)
alternativePaths: Alternative approaches suggested by Gemini
Session commands:
sessionCommand: Command to manage sessions ('save', 'load', 'getState')
sessionPath: Path to save or load the session file (required for 'save' and 'load' commands)
You should:
Start with a clear query and any relevant context
Let Gemini generate thoughts by not providing the 'thought' parameter
Review the generated thoughts and meta-commentary
Feel free to revise or branch thoughts as needed
Consider alternative paths suggested by Gemini
Only set next_thought_needed to false when truly done
Use session commands to save your progress and resume later
| Name | Required | Description | Default |
|---|---|---|---|
| query | Yes | The question or problem to analyze | |
| context | No | Additional context information | |
| thought | No | Your current thinking step (if empty, will be generated by Gemini) | |
| approach | No | Suggested approach to the problem | |
| branchId | No | Branch identifier | |
| isRevision | No | Whether this revises previous thinking | |
| sessionPath | No | Path to save or load the session file | |
| metaComments | No | Meta-commentary about the reasoning process | |
| thoughtNumber | Yes | Current thought number | |
| totalThoughts | Yes | Estimated total thoughts needed | |
| revisesThought | No | Which thought is being reconsidered | |
| sessionCommand | No | Command to manage sessions ('save', 'load', 'getState') | |
| confidenceLevel | No | Confidence level in the generated thought (0-1) | |
| alternativePaths | No | Alternative approaches suggested | |
| previousThoughts | No | Array of previous thoughts for context | |
| branchFromThought | No | Branching point thought number | |
| needsMoreThoughts | No | If more thoughts are needed | |
| nextThoughtNeeded | Yes | Whether another thought step is needed |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description fully covers behavioral aspects like meta-commentary, confidence levels, revision, branching, and session persistence. However, it omits potential costs or rate limits associated with the Gemini AI backend, slightly reducing transparency.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is well-organized with sections and bullet points, but it is overly verbose. Some redundancy exists between the 'when to use' list and 'key features'. Removing repetition would improve conciseness while retaining clarity.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the tool's complexity and lack of output schema, the description adequately explains usage flow and available actions. However, it does not explicitly specify the output structure (e.g., JSON containing generated thought, meta-comments, etc.), leaving minor ambiguity.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100%, so baseline is 3. The description adds extra context beyond schema descriptions, e.g., explaining that an empty 'thought' parameter will be AI-generated and that 'thought_number' can exceed initial estimates. This adds meaningful value.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool is for dynamic, reflective problem-solving via Gemini AI. It includes specific usage scenarios and features. However, with no sibling tools, differentiation is not assessed, but the purpose is clear and distinct from a generic 'think' tool.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description provides a detailed list of when to use the tool, including breaking down complex problems, planning, and analysis needing course correction. It does not explicitly mention when not to use it, but the extensive positive guidance compensates.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
v1.0.1geminithinkingScored across 1 tool
Only one tool exists, so there is no risk of confusion or overlap with other tools.
With a single tool, naming consistency is inherently perfect; the name 'geminithinking' clearly describes its purpose.
The server has only one tool, which feels thin for a domain that could benefit from separate tools for sessions, branching, or analysis management. However, the single tool is feature-rich.
The tool covers all expected functionalities for a deep thinking assistant: querying, context, revisions, branching, session persistence, and meta-commentary. No obvious gaps.
MCP server for generating rough-draft project plans from natural-language prompts.
MCP server for AI dialogue using various LLM models via AceDataCloud
MCP server unifying ERPs, CRMs, APIs and knowledge base for Claude, ChatGPT and Gemini.
MCP server for OpenAI API (chat completions, image generation, embeddings) via AceDataCloud
