latex-tools.online

Academic paper search for AI agents, served at latex-tools.online/mcp. Three corpora are available via the `source` argument: `arxiv` (default), `semanticscholar` (alias `s2`), and `openalex` (alias `oa`). Generic, source-agnostic tools: 1. `search_papers(query=..., source='arxiv', max_results=10, sort_by='relevance')` to find papers. For arxiv, `query` accepts plain text or field syntax like `ti:`, `au:`, `cat:cs.CL`, `abs:` + AND/OR. 1b. `search_all(query=..., max_results=10)` searches ALL three corpora at once, de-duplicates the same work across them (by DOI/title) and re-ranks with Reciprocal Rank Fusion; each hit carries `sources` (who found it) and an `ids` map you can hand to get_paper/read_paper. Use it as the default broad search; use `search_papers` when you want one specific corpus or arxiv field syntax. 1c. `search_medical(query=..., study_types='rct,meta-analysis,systematic-review', max_results=10)` for CLINICAL/biomedical questions: searches PubMed, filters by research type via Publication-Type tags and re-ranks by the evidence pyramid (meta-analysis / systematic review > RCT > cohort > ...), so real trials surface above high-cited reviews. Open-access full text is attached from Europe PMC. Prefer it over search_all for medicine; pass English keyword text (do NL/multilingual query understanding yourself). 2. `get_paper(paper_id=..., source='arxiv')` for one paper's full record. For s2, id accepts S2 id / `DOI:` / `ARXIV:` / `CorpusId:`. 3. `search_by_author(author=..., source='arxiv')` newest first. 4. `list_recent(category='cs.CL', source='arxiv')` latest in a category (arxiv code, or an S2 field of study). 5. `list_categories(source='arxiv')` category codes. 6. `read_paper(paper_id=..., format='markdown')` for the FULL text (arXiv): 'markdown' renders the body with formulas as $LaTeX$, 'html' returns the raw LaTeXML page, 'latex' returns the original manuscript source. 7. `list_paper_sources()` available corpora. Image → LaTeX (turn a formula or table image back into LaTeX, e.g. a figure cropped from a paper; no vision model needed on your side): • `recognize_formula(image_url=... or image_base64=...)` → LaTeX • `recognize_table(image_url=... or image_base64=...)` → LaTeX tabular • `list_ocr_models()` available OCR models Semantic Scholar capabilities (the full S2 API surface — citation graph, authors, recommendations, full-text snippets, bulk datasets): • `get_paper_citations` / `get_paper_references` / `get_paper_authors` • `match_paper_title` (exact title) / `autocomplete_papers` • `search_papers_bulk` (≤1000, sortable, token paging) / `get_papers_batch` • `search_authors` / `get_author` / `get_author_papers` / `get_authors_batch` • `search_snippets` (search inside paper full text) • `recommend_papers_for_paper` / `recommend_papers_from_examples` • `list_dataset_releases` / `get_dataset_release` / `get_dataset_download_links` / `get_dataset_diffs` OpenAlex capabilities (free CC0 all-field corpus, 316M works — citation graph, authors with h-index, institutions, topics, influence metrics): • `get_openalex_work` / `get_openalex_citations` / `get_openalex_references` • `search_openalex_authors` / `search_openalex_institutions` • `search_openalex_works` (filters: year range, open-access, min-citations, institution) • `get_openalex_trends` (publication-trend analytics) / `list_openalex_topics` Results are normalized across sources, so the same fields apply no matter which corpus is queried.