URL Context

Provides URL context analysis and optional Google Search grounding via Gemini API to summarize URLs with key facts and citations.

javascript

7

GitHub Stars

javascript

Language

6 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "nanameru-url-context-mcp": {
      "command": "npx",
      "args": [
        "@taiyokimura/url-context-mcp@latest"
      ],
      "env": {
        "GOOGLE_API_KEY": "sk-your-real-key"
      }
    }
  }
}

You can analyze and summarize the content of URLs using URL Context with optional grounding from Google Search. This MCP server leverages Google Gemini’s URL Context capability to produce concise summaries, key facts, and citation URLs, with the option to include Google Search results for broader context. It is designed for prompt-only orchestration, letting you decide when to search or scrape through your instruction text.

How to use

You run this MCP server through an MCP client or integration that supports standard MCP commands. The server exposes a single primary tool named analyze_urls. You provide one or more URLs, an instruction in your preferred language, and optional flags to control behavior. The tool will return a concise summary, key facts, and a list of cited URLs. You can enable grounding with Google Search to enrich results.

Practical usage patterns include providing a single URL or multiple URLs with clear instructions. For example, you can request a Japanese summary and key facts for a specific article, or you can request iterative search-and-scrape with up to five rounds if the coverage is not sufficient. Use use_google_search to toggle grounding with Google Search.

How to install

Prerequisites you need before running the server are Node.js 18+ and a Gemini API key. You also need to have a compatible MCP client capable of invoking the analyze_urls tool.

Install and run the MCP server using the published package. You can execute it directly without cloning the repository by using a package manager command (see below).

Step-by-step commands you can follow:

Ensure you have Node.js 18 or newer installed
Run the MCP server via the package runner
Provide your Google Gemini API key as an environment variable when starting the server

Configuration and usage notes

Environment variable you must set: GOOGLE_API_KEY with your Google Gemini API key. This key is required for both URL Context and optional Google Search grounding.

Available tool:

analyze_urls — inputs: urls (string or string[]; 1-20 total), instruction (string), model (string; default is gemini-2.5-flash), use_google_search (boolean; default false). Grounding with Google Search is enabled by setting use_google_search to true.

Example invocation follows the MCP tool call format and demonstrates enabling Google Search grounding.

Troubleshooting

401 auth errors usually indicate the Gemini API key is missing or invalid. Double-check that GOOGLE_API_KEY is set correctly in your environment where the MCP server runs.

Ensure you are using Node 18+ and that your MCP client is compatible with this server. If you are running locally, use the standard npx-based command to start the server.

Notes and examples

The server supports iterative search-and-scrape up to five rounds when you enable Google Search grounding. If the coverage from URL Context alone is insufficient, the server can re-run searches and refine results while keeping the same instruction.

Available tools

analyze_urls

Analyze one or more URLs to produce a concise summary, key facts, and cited URLs. Optionally ground results with Google Search to enrich context.