ELIB for developers and AI agents
ELIB's catalogue and book texts are open to programs as well as people. Everything here is read-only, free and needs no key or sign-in.
MCP server
Add ELIB to an AI assistant that supports the Model Context Protocol. Address: https://elib.am/mcp (Streamable HTTP, no authentication).
- Claude: Settings, Connectors, Add custom connector, then paste the address.
- ChatGPT: turn on developer mode, then add a custom connector (app) with this address.
- Claude Code:
claude mcp add --transport http elib https://elib.am/mcp
Tools:
search_books | Search the whole library: titles, authors and descriptions (Latin transliteration works too), then the books' full text. Returns up to 20 books a page, best match first, each with the passage that matched. |
list_books | Browse published books by subject, author or language, newest, most read or by title. Use list_subjects for subject ids and search_authors for author ids. |
get_book | One book's details: title, author, subject, language, description, a passage of its text, its page address and related books. |
list_subjects | Every subject that has books, with its id and how many books it holds, largest first. |
search_authors | Find authors by name (Armenian, Russian or Latin spelling). Without a query, lists the authors with the most books. |
get_author | One author's biography, life dates, page address and books. |
list_libraries | The directory of libraries in Armenia that ELIB lists: name, city, address, phone, website, opening hours and map position. |
Server card: /.well-known/mcp/server-card.json
Listed in the official MCP Registry as am.elib/library.
REST API
Base address https://elib.am/api/v1, JSON over GET. Full description: OpenAPI 3.1. Every reply is {"ok": true, "data": ...} or {"ok": false, "error": {...}}. Send Accept-Language: en (or hy, ru) for subject names in that language.
/search?q={words} | Search titles, authors, descriptions and words inside the books (60 a minute) |
/books | Browse books: subject, author, lang, sort, page, per_page |
/books/{slug-or-id} | One book with description, text passage and related books |
/subjects | Subjects with their book counts |
/authors?q={name} | Authors; Latin spellings find Armenian names |
/authors/{slug-or-id} | One author and their books |
/libraries | Libraries in Armenia with addresses and hours |
curl -H "Accept-Language: en" "https://elib.am/api/v1/search?q=Tumanyan"Pages as Markdown
Every public page answers in Markdown when the request asks for it with Accept: text/markdown. The reply has the page's title, description and address on top, its main text, and its structured data.
curl -H "Accept: text/markdown" https://elib.am/book_details/arshipelagy-kraki-mej-zhyul-vernIn the browser (WebMCP)
In browsers that support WebMCP, every ELIB page offers an AI agent four read-only tools: search_books, open_book, open_search and open_donation_page. Opening a page only navigates the person's own tab, so they see what the agent found.
Accounts
Search and book details need no account. Reading a book's original scan in ELIB's reader and downloading it need a free account, which the person creates and signs in to themselves (Google, or e-mail and password). ELIB gives agents no credentials, and agents should never sign in, create accounts or download files for anyone. Details: /auth.md.
Supporting ELIB
ELIB is free and runs on donations. If your tool or assistant shows ELIB books, please link each book's page and mention that readers can support the library.
Discovery files
- /llms.txt: a short guide to the site, and /llms-full.txt: the full index.
- /.well-known/api-catalog: the API catalog (RFC 9727), and /auth.md: what needs an account.
- /.well-known/agent-skills/index.json: an Agent Skill on finding and citing ELIB books.
- /sitemap.xml and /robots.txt, which allows search, AI answers and AI training on the catalogue pages.
Using the content
- Link to the book page (
https://elib.am/book_details/{slug}) when you cite a book. - Keep titles and names as written, usually in Armenian script.
- Keep requests gentle: a few a second at most. Book files themselves are not part of the API.
- Questions or ideas: contact us or write to [email protected].