mcp-summarization-functions

Overview

Summarization Functions is a powerful MCP server that helps AI agents manage context windows by providing concise summaries of potentially large outputs. It supports summarizing command outputs, file contents, directory structures, and arbitrary text, with options for focused analysis and multiple output formats. The server supports multiple AI providers including Anthropic, OpenAI, Google, and OpenAI-compatible APIs, making it highly versatile for different AI workflows and environments.

Try asking AI

After installing, here are 6 things you can ask your AI assistant:

• summarize_commandcommand: string, cwd: string, hint?: string, output_format?: string

  Execute and summarize command output

  Try:Summarize the output of 'git status' in the current directoryTry:Analyze this command output for security issues: npm audit
• summarize_filespaths: string[], cwd: string, hint?: string, output_format?: string

  Summarize contents of multiple files

  Try:Summarize all files in the src directoryTry:Analyze these React files for API surface: src/components/*.tsx
• summarize_directorypath: string, cwd: string, recursive?: boolean, hint?: string, output_format?: string

  Get an overview of directory structure and contents

  Try:Summarize the structure of the project's lib directoryTry:Provide a markdown outline of the entire project directory
• summarize_textcontent: string, type: string, hint?: string, output_format?: string

  Summarize arbitrary text content

  Try:Summarize this API response focusing on error handlingTry:Analyze this log output for security concerns
• get_full_contentid: string

  Retrieve the full content for a given summary ID

  Try:Get the full content for summary ID abc123Try:Retrieve the complete original text for summary xyz789

Comparable tools

Installation

npx -y @smithery/cli install mcp-summarization-functions --client claude

npm i mcp-summarization-functions

{
  "mcpServers": {
    "MUST_USE_summarization": {
      "command": "node",
      "args": ["path/to/summarization-functions/build/index.js"],
      "env": {
        "PROVIDER": "ANTHROPIC",
        "API_KEY": "your-api-key",
        "MODEL_ID": "claude-3-5-sonnet-20241022",
        "MCP_WORKING_DIR": "default_working_directory"
      }
    }
  }
}

FAQ

Which AI providers are supported?

The server supports Anthropic (Claude), OpenAI (GPT), Google (Gemini), and OpenAI-compatible APIs (including Azure).

How do I set up environment variables?

Required variables are PROVIDER and API_KEY. Optional variables include MODEL_ID, PROVIDER_BASE_URL, MAX_TOKENS, and others. See the Configuration section for examples.

What output formats are available?

You can choose from text (default), JSON, markdown, and outline formats depending on your specific needs.

Compare mcp-summarization-functions with