# 0xshellming/mcp-summarizer [Health: Active]

**Category:** 🧠 Knowledge & Memory  
**Repository:** https://github.com/0xshellming/mcp-summarizer  
**GitHub Stars:** 167  
**Views:** 6  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/0xshellming-mcp-summarizer

## Description
AI Summarization MCP Server, Support for multiple content types: Plain text, Web pages, PDF documents, EPUB books, HTML content

## Tools
Capabilities this server exposes over MCP:

- **content**
- **type**
- **maxLength**
- **language**
- **focus**
- **style**

## Claude Desktop Quick Installation
Heuristic fallback — verify the package name and runner against the repository README before running it. Uses `npx` (confidence: low):

```json
"mcpServers": {
  "mcp-summarizer": {
    "command": "npx",
    "args": ["-y","0xshellming-mcp-summarizer"]
  }
}
```

## Documentation

## What 0xshellming/mcp-summarizer MCP server does

The 0xshellming/mcp-summarizer MCP server exposes content summarization through the Model Context Protocol. Its main `summarize` tool accepts source material and returns a generated summary. Supported source categories described by the project include plain text, web pages, PDF documents, EPUB books, and HTML content.

The input can be a text string, a web URL, base64-encoded PDF data, or EPUB file content. The tool uses a `type` value to distinguish the source format. The documented type values are `text`, `url`, `pdf`, and `epub`; HTML is listed among the supported content formats, but the excerpt does not provide a separate HTML type value.

## How it works

Summarization is performed with Google's Gemini 1.5 Pro model through the 3MinTop service. The caller can set `maxLength` to control the maximum summary size, with a documented default of 200 characters. `language` selects the output language and defaults to English. An optional `focus` value directs attention to a particular aspect of the source, while `style` can request a concise summary, a detailed summary, or bullet points.

For example, an agent can pass a URL with `type: "url"` to summarize a web page, or provide base64 PDF content with `type: "pdf"`. The server also includes a dynamic greeting resource using URIs in the form `greeting://{name}`. That resource demonstrates MCP resource behavior rather than content summarization.

The 0xshellming/mcp-summarizer MCP server is therefore suited to agents that need to reduce reading material before presenting it, extract a selected theme from a document, or produce summaries in a requested language and format. The provided material does not describe chunking, citation output, persistent storage, or batch processing.

## Setup and configuration

The repository's documented setup is local and source-based. Clone the repository, install dependencies with `pnpm install`, compile it with `pnpm run build`, and start it with `pnpm start`. Development work can use `pnpm run dev`, which runs the TypeScript compiler in watch mode. Server behavior is customized in `src/index.ts`.

A desktop MCP client can launch the built server with Node using `dist/index.js` as the argument. The README shows an absolute path is required in that client configuration, so the path must be supplied by the person installing it. No environment variable names or credential configuration are specified in the provided material.

## Tools and capabilities

- Summarize text, URLs, PDFs, EPUB content, and listed HTML content.
- Set a maximum summary length.
- Request a target language.
- Focus the summary on a specified aspect.
- Choose concise, detailed, or bullet-point output.
- Access the `greeting://{name}` demonstration resource.

## Limitations and notes

The README identifies Gemini 1.5 Pro and 3MinTop as the summarization backend, but does not explain service limits, supported file-size limits, error handling, network requirements, or credential setup. It also does not identify named desktop clients beyond referring generally to a desktop app. The project is licensed under MIT.

The 0xshellming/mcp-summarizer MCP server's documented installation flow builds and starts the repository locally; the supplied material does not identify a published npm, PyPI, Docker, or hosted MCP package.

_Full upstream README: https://allmcps.com/mcp/0xshellming-mcp-summarizer/readme_