annas-mcp

Anna's Archive MCP Server (and CLI Tool)

An MCP server and CLI tool for searching and downloading documents (books and articles) from Anna's Archive, with optional automatic selection of a mirror reported as healthy by SLUM.

[!NOTE] Notwithstanding prevailing public sentiment regarding Anna's Archive, the platform serves as a comprehensive repository for automated retrieval of documents released under permissive licensing frameworks (including Creative Commons publications and public domain materials). This software does not endorse unauthorized acquisition of copyrighted content and should be regarded solely as a utility. Users are urged to respect the intellectual property rights of authors and acknowledge the considerable effort invested in document creation.

[!WARNING] Please refer to Mirror Selection if any of the links lead to a non-functional Anna's Archive server.

Requirements

Search works without any required environment variables.

Downloads require:

• A donation to Anna's Archive, which grants JSON API access
• An API key
• An MCP client, such as Claude Desktop, if using the project as an MCP server

Setup

Download the appropriate binary from the GitHub Releases section.

If you plan to use the tool for its MCP server functionality, you need to integrate it into your MCP client. If you are using Claude Desktop, please consider the following example configuration:

"anna-mcp": {
    "command": "/Users/iosifache/Downloads/annas-mcp",
    "args": ["mcp"],
    "env": {
        "ANNAS_SECRET_KEY": "feedfacecafebeef",
        "ANNAS_DOWNLOAD_PATH": "/Users/iosifache/Downloads",
        "ANNAS_BASE_URL": "annas-archive.gl"
    }
}

Configuration

The tool can be configured with environment variables. These variables can also be stored in an .env file in the folder containing the binary.

Download Environment Variables

Downloads use:

• ANNAS_SECRET_KEY: The Anna's Archive API key.
• ANNAS_DOWNLOAD_PATH: The path where the documents should be downloaded.

Mirror Selection

Anna's Archive has multiple mirrors, and their availability can change over time. By default, this project uses ANNAS_BASE_URL, or annas-archive.gl when ANNAS_BASE_URL is not set.

Optionally, you can set:

• ANNAS_BASE_URL: The Anna mirror to use (defaults to annas-archive.gl). When automatic mirror discovery is enabled, this becomes the fallback mirror.
• ANNAS_AUTO_BASE_URL: Set to true to discover the best available Anna mirror automatically from SLUM.

Automatic discovery is opt-in: when ANNAS_AUTO_BASE_URL=true, the tool reads the public status page, ranks discovered Anna mirror candidates by recent health and latency, probes them locally, and uses the best reachable mirror. If discovery or probing fails, the tool falls back to ANNAS_BASE_URL, then to the built-in default mirror.

Timeouts

HTTP requests default to a 1 hour timeout. For CLI usage, override this with --timeout, for example:

annas-mcp --timeout 1h book-download abc123def456 "my-book.pdf"

For MCP usage, tools accept an optional timeout_seconds parameter, for example 3600 for 1 hour.

Demo

As an MCP Server

As a CLI Tool

Available Operations

Operation	MCP Tool	CLI Command	Example
Search for books by title, author, or topic	book_search	book-search	book-search "machine learning python"
Download a book by its MD5 hash	book_download	book-download	book-download abc123def456 "my-book.pdf"
Search for articles by DOI or keywords	article_search	article-search	article-search "10.1038/nature12345" or article-search "neural networks"
Download an article by its DOI	article_download	article-download	article-download "10.1038/nature12345"