Skip to main content
Glama
shgsousa

serper-search

by shgsousa
OverviewSchemaSourceRelated ServersDiscussionsAdmin
JavaScript
Remote

Serper Search MCP Server

A Model Context Protocol (MCP) server that enables web searching using the Serper API for Google search results.

Features

  • Search the web using Serper API for Google search results

  • Requires Serper API key for authentication

  • Returns structured results with titles, URLs, and descriptions

  • Fetches and includes actual web page content for each result

  • Configurable number of results per search

  • Supports streamable-http transport for LibreChat integration

  • Docker containerization support

  • Health checks and monitoring

  • Built-in rate limiting to respect API limits

Related MCP server: Serper MCP Server

Installation

  1. Clone or download this repository

  2. Install dependencies:

npm install

  1. Build the server:

npm run build

Usage Modes

Local Development (Stdio Mode)

For local development and direct MCP client integration:

npm start

Add the server to your MCP configuration:

For VSCode (Claude Dev Extension):

{
  "mcpServers": {
    "web-search": {
      "command": "node",
      "args": ["/path/to/web-search/build/index.js"]
    }
  }
}

For Claude Desktop:

{
  "mcpServers": {
    "web-search": {
      "command": "node",
      "args": ["/path/to/web-search/build/index.js"]
    }
  }
}

HTTP Mode (Container-Ready)

For containerized deployments or LibreChat integration:

# Start in HTTP mode
npm run start:http
# or
node build/index.js --http
# or set environment variable
MCP_HTTP_MODE=true npm start

The server will expose:

  • MCP endpoint: http://localhost:3000/mcp (for JSON-RPC 2.0 requests)

  • Health endpoint: http://localhost:3000/health

  • Health check: http://localhost:3000/health

Docker Deployment

Using Docker directly:

# Build the image
npm run docker:build

# Run the container
npm run docker:run

Or manually:

docker build -t web-search-mcp .
docker run -p 3000:3000 -e MCP_HTTP_MODE=true web-search-mcp

Using Docker Compose:

# Start the service
npm run docker:up

# View logs
npm run docker:logs

# Stop the service
npm run docker:down

Container Configuration

Environment variables:

  • MCP_HTTP_MODE: Set to true to enable HTTP/SSE mode

  • PORT: Port number (default: 3000)

  • SEARCH_RATE_LIMIT_MS: Minimum milliseconds between search requests (default: 500)

The container includes:

  • Health checks

  • Non-root user execution

  • CORS support

  • Automatic restart policies

Rate Limiting

The server includes built-in rate limiting to be respectful to Google's servers:

  • Default: Minimum 500ms between search requests

  • Configurable: Set SEARCH_RATE_LIMIT_MS environment variable

  • Automatic: If requests come in faster than the limit, the server will automatically wait

  • Logging: Rate limiting events are logged to stderr

Example with custom rate limit:

# Set 1 second minimum between searches
SEARCH_RATE_LIMIT_MS=1000 npm run start:http

API Reference

Tool: search

Parameters:

{
  "query": string,    // The search query
  "limit": number,    // Optional: Number of results to return (default: 5, max: 10)
  "maxContentLength": number  // Optional: Max length of content to extract (default: 50000)
}

Tool: fetch_page_content

Parameters:

{
  "url": string,    // The URL of the web page to fetch
  "maxContentLength": number  // Optional: Max length of content to extract (default: 50000)
}

HTTP/SSE API

Health Check

GET /health

Returns:

{
  "status": "ok",
  "service": "web-search-mcp"
}

SSE Connection

GET /sse

Establishes Server-Sent Events connection for real-time communication.

Message Endpoint

POST /message
Content-Type: application/json

{
  "jsonrpc": "2.0",
  "id": "unique-id",
  "method": "tools/call",
  "params": {
    "name": "search",
    "arguments": {
      "query": "your search query",
      "limit": 5
    }
  }
}

Testing

A test client is included (test-client.html) for testing the HTTP/SSE endpoint. Open it in a browser and ensure the server is running in HTTP mode.

Example Usage

MCP Client (stdio mode):

use_mcp_tool({
  server_name: "web-search",
  tool_name: "search",
  arguments: {
    query: "your search query",
    limit: 3
  }
})

HTTP API (container mode):

const response = await fetch('http://localhost:3000/message', {
  method: 'POST',
  headers: { 'Content-Type': 'application/json' },
  body: JSON.stringify({
    jsonrpc: "2.0",
    id: "1",
    method: "tools/call",
    params: {
      name: "search",
      arguments: { query: "Model Context Protocol", limit: 5 }
    }
  })
});

Example response:

{
  "jsonrpc": "2.0",
  "id": "1",
  "result": {
    "content": [
      {
        "type": "text",
        "text": "[{\"title\":\"Example Result\",\"url\":\"https://example.com\",\"description\":\"Description...\"}]"
      }
    ]
  }
}

Fetch Page Content Tool

HTTP API Example

{
  "jsonrpc": "2.0",
  "id": "1",
  "method": "tools/call",
  "params": {
    "name": "fetch_page_content",
    "arguments": {
      "url": "https://en.wikipedia.org/wiki/Model_Context_Protocol",
      "maxContentLength": 10000
    }
  }
}

Example Response

{
  "jsonrpc": "2.0",
  "id": "1",
  "result": {
    "content": [
      {
        "type": "text",
        "text": "[Cleaned web page content here...]"
      }
    ]
  }
}

Limitations

Since this tool uses web scraping of Google search results, there are some important limitations to be aware of:

  1. Rate Limiting: Google may temporarily block requests if too many searches are performed in a short time. To avoid this:

    • Keep searches to a reasonable frequency

    • Use the limit parameter judiciously

    • Consider implementing delays between searches if needed

  2. Result Accuracy:

    • The tool relies on Google's HTML structure, which may change

    • Some results might be missing descriptions or other metadata

    • Complex search operators may not work as expected

  3. Legal Considerations:

    • This tool is intended for personal use

    • Respect Google's terms of service

    • Consider implementing appropriate rate limiting for your use case

Contributing

Feel free to submit issues and enhancement requests!

License

This project is licensed under the MIT License. See the LICENSE file for details.

Available Tools

2 tools
fetch_page_contentC

Fetch and clean the main content of a web page, with a configurable content length limit

ParametersJSON Schema
NameRequiredDescriptionDefault
urlYesThe URL of the web page to fetch
maxContentLengthNoMaximum length of content to extract (default: 50000 characters)

TDQS

C2.9/5.0
Behavior2/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

No annotations are provided, so the description carries the full burden of behavioral disclosure. It mentions 'cleaning' and a configurable length limit, but does not disclose return format, error handling, redirects, authentication requirements, or what happens with non-HTML content. This is insufficient for a fetch tool.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is a single concise sentence that immediately conveys the core action and configurable option. It is front-loaded, contains no redundant words, and every element is relevant.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness2/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

With no output schema and no annotations, the description should clarify what the returned content looks like, how the max length is applied, and behavior on failures. The description is too minimal to be complete for a fetch operation, leaving significant gaps for an agent to act reliably.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100% for both parameters, so the schema fully documents 'url' and 'maxContentLength'. The description's mention of a 'configurable content length limit' aligns with maxContentLength but adds little beyond what the schema already conveys, warranting the baseline score of 3.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the tool fetches and cleans the main content of a web page, with a specific verb and resource. It implicitly distinguishes from the sibling 'search' tool by focusing on page content extraction rather than search, though it does not explicitly name the alternative.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines2/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description provides no guidance on when to use this tool versus the sibling 'search' tool. There is no mention of scenarios, prerequisites, or exclusions, leaving the agent to infer usage from the generic purpose statement.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

  1. 2 tool updatesv0.1.0
    • First observedfetch_page_content
    • First observedsearch
Deploy Server
RayrunBuild, deploy, and operate MCP servers without becoming an infrastructure team.ray.runAd

TDQS

B3.3/5.0

Scored across 2 tools

Disambiguation4/5

The two tools are distinct in primary purpose: search is for query-based discovery, fetch_page_content is for retrieving a specific URL. However, the search tool also fetches content from result pages, creating slight overlap that could cause confusion, but descriptions help clarify.

Naming Consistency4/5

Both tool names use an imperative verb style, but 'search' is a single verb while 'fetch_page_content' is a compound verb_noun. This is a minor inconsistency, yet the names remain clear and predictable.

Tool Count3/5

With only two tools, the server feels thin for a search-focused MCP. They cover the core steps of search and content retrieval, but the count is at the borderline where the toolset could be perceived as minimal.

Completeness4/5

The domain of web search and content fetching is adequately covered: search discovers pages and fetches their content, while fetch_page_content handles arbitrary URLs. Missing features like search customization or pagination are minor and do not create dead ends.

Maintenance

ActivityInactive
ResponsivenessNo issues

Resources

Related MCP Connectors

Related MCP Servers