What Anna's Archive MCP server does
Anna's Archive MCP server exposes member-authenticated access to Anna's Archive through MCP tools and a matching command-line interface. It is intended for finding books, textbooks, manuals, standards, journal articles, and related catalog records, then downloading a selected file.
Book searches can use title, author, or topic and can be narrowed by format, language, source, or year. Article searches accept keywords, authors, or years, and can also look up an article by DOI. Optional file activity counts can be included in both search types. Metadata searches query catalogs including Open Library, WorldCat, and ISBNdb to help identify editions and ISBNs.
The remaining operations support inspection and retrieval. book_details reports identifiers, edition information, descriptions, and reported file problems. account_status shows membership information and the number of fast downloads remaining. Book downloads accept a file hash or Anna's Archive book-page URL, while article downloads accept a DOI or hash.
How it works
The client starts the MCP process with Node.js, and the server signs in using the member's secret key before making its first request. It renews the session when needed and saves the session between runs, so browser cookies do not normally need to be copied. If key-based sign-in fails, an aa_account_id2 browser cookie can be supplied as a fallback.
Downloads are checked against the archive's MD5 hash. The server can select a working official mirror automatically, or use a configured base URL. Files are written to a configurable local folder, which defaults to an annas-archive directory in the user's home folder.
The package also includes annas-cli, which supports the same operations from a terminal. Go applications can use the github.com/SokolskyNikita/annas-mcp/anna SDK for search, metadata, account, and verified-download operations; the SDK requires Go 1.26 or newer.
Setup and configuration
Install or run the package with Node.js 18 or newer. Set ANNAS_SECRET_KEY to the secret key from the Anna's Archive account page, then register annas-mcp in a supported MCP client or run annas-cli directly. The README provides configuration examples for Claude Desktop, Claude Code, Cursor, Codex, Gemini CLI, and Docker.
ANNAS_DOWNLOAD_PATH changes the destination for files. ANNAS_ACCOUNT_COOKIE supplies the optional browser session cookie, while ANNAS_BASE_URL selects a mirror to try first. Set ANNAS_AUTO_BASE_URL=false to prevent automatic mirror selection. Docker users should mount the host download directory at /downloads.
A paid Anna's Archive membership is required. The Brilliant Bookworm tier supports downloads, while searching through MCP requires Lucky Librarian or a higher tier. The documentation recommends the latter when the agent must both search and download.
Limitations and notes
Anna's Archive MCP server depends on an active membership and a valid sign-in credential. Without a paid membership, neither search nor download is available through the documented workflow. With the basic paid tier, users may need to provide each book download URL manually because MCP search is restricted to the higher membership level.
The service uses the archive's English interface; language-specific mirror versions are refused. Keep the secret key and any account cookie private. Downloads should be limited to material the user is entitled to obtain under applicable laws and terms.