広告掲載
MCP.so
ログイン

Deep Research MCP Server

@ssdeanx

リポジトリホームページ

基本情報

カテゴリ

データと分析

ライセンス

MIT

トランスポート

stdio

作者

ssdeanx

Deep Research MCP Server について

MCP Deep Research Server using Gemini creating a Research AI Agent

#research#ai#research-tool#research-paper#research-and-development#ai-agents#ai-research#ai-researcher#airesearch#aiagent#ai-research-assistant#mcp-server

Deep Research MCP Server の設定

以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。

{
  "mcpServers": {
    "deep-research-mcp-server": {
      "command": "node",
      "args": [
        "--env-file",
        ".env.local",
        "dist/mcp-server.js"
      ]
    }
  }
}

Deep Research MCP Server のツール

ツール一覧はまだありません

この掲載には表示できるツール一覧がまだありません。サーバーにツールがないという意味ではありません。対応機能はプロジェクトのドキュメントをご確認ください。

ツールは README から自動的に抽出されます。メンテナーは ## Tools という見出しの下に記載することで、このタブに反映できます。

概要

What is Deep-research?

Deep-research is an AI-powered research assistant that runs as a Model Context Protocol (MCP) server. It uses Google Gemini 2.5 Flash with Google Search Grounding to perform iterative, deep research on user queries, producing professional Markdown reports—all without requiring web scraping dependencies.

How to use Deep-research?

To run as an MCP tool, start the server with node --env-file .env.local dist/mcp-server.js and invoke the deep-research tool with parameters such as query, depth, and breadth. Alternatively, use the standalone CLI by running npm run start "your research query". The server is configured via environment variables like GEMINI_API_KEY, GEMINI_MODEL, and CONCURRENCY_LIMIT.

Key features of Deep-research

  • MCP server/tool integration for seamless agent workflows
  • Gemini 2.5 Flash pipeline with structured JSON outputs
  • Iterative deep dive with query refinement and context carryover
  • Configurable depth and breadth for exploration scope
  • Semantic and recursive splitting for robust content analysis
  • Batching and LRU caching for concurrency-limited performance
  • Generates professional Markdown reports with abstract, methodology, and references

Use cases of Deep-research

  • Conducting iterative literature reviews on emerging topics
  • Generating structured research reports for multi-agent systems
  • Exploring multi-faceted questions with controlled depth and breadth
  • Integrating as a research tool within MCP-compatible agent clients

よくある質問

1

What are the system requirements for Deep-research?

Node.js v22.x and a Google Gemini API key are required. The server runs on the Gemini 2.5 Flash model.

2

How do I configure Deep-research?

Configuration is done through environment variables set in a `.env.local` file. Required: `GEMINI_API_KEY`. Optional: `GEMINI_MODEL`, `GEMINI_MAX_OUTPUT_TOKENS`, `CONCURRENCY_LIMIT`, and flags to enable Gemini tools (Google Search Grounding, Code Execution, Functions).

3

Can I use Deep-research without an MCP client?

Yes, Deep-research can be run as a standalone CLI tool using `npm run start "query"`, bypassing the MCP server.

4

Does Deep-research require web scraping dependencies?

No, the pipeline uses Google Search Grounding via Gemini’s tool use, so no web scraping library is needed.

5

What transport and authentication does the MCP server use?

The MCP server uses the standard MCP transport. The `GEMINI_API_KEY` must be supplied to the server process as an environment variable for authentication.

コメント

「データと分析」の他のコンテンツ

AskWatch Google Search Console MCP

AskWatch

Ask Google Search Console in plain language from Claude, ChatGPT, Cursor or Claude Code. Connect your Google account once and AskWatch issues a personal link: no Cloud project, no credentials.json, nothing to run or host

Draxlr

Vivek

Draxlr is a remote MCP server that connects AI assistants to your SQL databases. Explore schemas, run read-only queries, and work with your saved queries and dashboards over MCP. Hosted at https://api.draxlr.com with OAu

Runzumi

Runzumi

Explore GitHub Actions job performance, compute costs and failures with Runzumi’s remote MCP server. Connect securely with OAuth and query the workspaces and repositories you can access.

Instagram MCP Server

HasData

Read public Instagram profiles and post feeds by handle in Claude, Cursor, and other MCP clients. Get follower counts, captions, hashtags, mentions, and engagement counts as structured JSON.

23

Tracetify — Competitor Ads, Growth Traces & Search Console

tracetify

17 tools for marketing and SEO research inside Claude Code, Cursor, Codex or ChatGPT: the ads any domain runs on Meta and Google, how a product grew from 12 public sources with every claim linked, your own Search Console

1
広告掲載