Source Library

skills: {'id': 'search_library', 'name': 'search_library', 'description': 'Find BOOKS matching a topic. Searches titles, authors, subjects, and (as a secondary signal) translated text. Use this when you want a list of works on a subject. For locating specific passages inside books, use search_translations instead. ORIENTATION HINT: when the user has named a specific author or work, call get_book directly (or list_books to discover the ID) — the AI-generated book summary + chapter outline is often the right first answer and saves repeated passage hunting. Query tips: single distinctive words or short phrases work best ("memory palace", "ouroboros"); quoted phrases match exactly. Each result includes total_matches (full count) + returned (this page) + offset for pagination.', 'tags': [], 'examples': None, 'input_modes': None, 'output_modes': None}, {'id': 'search_translations', 'name': 'search_translations', 'description': 'Find specific PASSAGES inside books — returns page-level snippets with citation URLs. Use this when you want a quote or evidence on a topic across the whole library. ORIENTATION HINT: if the user has named a specific author or work, prefer get_book (returns a summary + chapter outline) over passage hunting — every book in the corpus has an AI-generated summary that is usually the right first read. Use search_translations when sweeping across many books for evidence of a theme. For finding which BOOKS cover a topic, use search_library. Query tips: single distinctive terms ("memory palace", "wax tablet") work best; multi-word natural-English queries ("unity of the intellect") may return fewer results because matching is term-based, not phrase-based. Each snippet has a snippet_type — "translation"/"ocr" means it is a verbatim extract from the source text; "summary" means it is AI-generated description (do not quote those as the author\'s words). Response includes total_matches, returned, and offset for pagination. Cross-cultural tip: for pre-modern or non-Western topics, search source-tradition vocabulary rather than modern English terms — e.g. for seminal economy search "jing" or "bindu" or "istimnāʾ", not "semen retention"; for female homoeroticism search "tribade" or "sahq", not "lesbian". The corpus is indexed via period translations that use tradition-internal terminology.', 'tags': [], 'examples': None, 'input_modes': None, 'output_modes': None}, {'id': 'search_concept', 'name': 'search_concept', 'description': 'Conceptual / semantic passage search across the whole library. Use when the modern term won\'t literally appear in historical texts — e.g. "distributed cognition" maps to passages about active intellect, art of memory, wax tablet metaphors; "social contract" maps to pre-Hobbesian discussions of consent and authority. Ranks passages by cosine similarity on Gemini embeddings (768d), so paraphrases and conceptually adjacent phrasings match even when no keyword overlaps. ORIENTATION HINT: if the user named a specific author or work, prefer get_book (returns the book\'s AI summary + chapter outline) — semantic search is expensive and best reserved for cross-corpus discovery. Prefer search_translations for literal phrases or distinctive single terms; use search_concept when the concept matters more than the wording. Similarity calibration: 0.70+ is a strong match, 0.55–0.70 is worth reading but verify, below 0.55 is mostly conceptual drift. Set max_per_book to diversify results across many books rather than cluster on one source. Each passage carries a snippet_type — quote only "translation" snippets, never "summary". Cross-cultural tip: for pre-modern or non-Western topics, also try source-tradition vocabulary — e.g. for seminal economy try "jing preservation" or "bindu yoga" or "istimnāʾ"; for masturbation try "mollities" (Latin) or "hastamaithuna" (Sanskrit) or "shouyin" (Chinese). The corpus is indexed via period translations that use tradition-internal terminology, so adjacent/euphemistic terms often surface material that modern English keywords miss.', 'tags': [], 'examples': None, 'input_modes': None, 'output_modes': None}, {'id': 'search_within_book', 'name': 'search_within_book', 'description': 'Deep-dive inside a single book. Runs Atlas keyword search AND scoped semantic search in parallel against that book\'s pages, then merges results — so this works for both literal terms ("ouroboros") and conceptual queries ("the marriage of opposites"). Typical workflow: use search_library or search_concept to find a candidate book; then call this with that book_id to surface every relevant page. Faster than re-searching globally because it\'s scoped to one book\'s 100-500 pages. Returns OCR and translation snippets with page numbers, ready to cite.', 'tags': [], 'examples': None, 'input_modes': None, 'output_modes': None}, {'id': 'list_books', 'name': 'list_books', 'description': 'Browse the catalog by metadata — filter by author/title fragment, language, category, or translation recency. Returns books with title, author, language, year, and translation progress. Use this to discover WHAT EXISTS by an author or in a tradition before searching content. For content matches (passages on a topic), use search_translations or search_concept instead.', 'tags': [], 'examples': None, 'input_modes': None, 'output_modes': None}, {'id': 'get_book', 'name': 'get_book', 'description': "Get a book's AI-generated summary, chapter list, edition metadata, DOI, and page counts. THIS IS THE RIGHT FIRST CALL whenever the user has named a specific author or work — the summary is typically a multi-paragraph orientation covering the book's argument, structure, and significance, often answering the question without any further searching. Pair with get_book_text to read selected chapters, or search_within_book to locate passages inside it.", 'tags': [], 'examples': None, 'input_modes': None, 'output_modes': None}, {'id': 'get_book_text', 'name': 'get_book_text', 'description': "Read a book's text. Preferred: use the chapter param to read one chapter at a time (includes [Page N] markers for citation) — call get_book first to get the chapter list. Alternatively, use from/to for explicit page ranges (e.g. from=1 to=50). TRUNCATION: the response always includes truncated: true/false. When truncated=true, the truncation_note field gives the exact next from/to values to call — this means content was cut short by a page-budget limit, NOT that the book ended. An AI agent MUST NOT infer end-of-book from pages_returned alone; check truncated first. Budget limits apply to anonymous callers (~50 pages per 24h); sign in at sourcelibrary.org/auth/signin or get an API key at sourcelibrary.org/developers for higher limits.", 'tags': [], 'examples': None, 'input_modes': None, 'output_modes': None}, {'id': 'get_quote', 'name': 'get_quote', 'description': 'Get exact text of a single page for quoting, with citation URL. ALWAYS use before putting text in quotation marks.', 'tags': [], 'examples': None, 'input_modes': None, 'output_modes': None}; uptime_30d 1.0%; p95 777.5ms; conformance: pass

How to connect

https://sourcelibrary.org/api/mcp

curl -X POST https://sourcelibrary.org/api/mcp \
  -H 'Content-Type: application/json' \
  -H 'Accept: application/json, text/event-stream' \
  -d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{}}'