← ClaudeAtlas

paper-searchlisted

Search, download, and extract text from academic papers across arXiv, PubMed, Semantic Scholar, Crossref, OpenAlex, and other scholarly sources. Use when the user asks to find literature, compare search results, resolve a DOI or paper identifier, download a paper PDF, or read paper text through the installed paper-search CLI.
alfredzhang98/research-obsidian-skill · ★ 1 · AI & Automation · score 64
Install: claude install-skill alfredzhang98/research-obsidian-skill
# Paper Search Use the `paper-search` CLI installed from the attributed `openags/paper-search-mcp` dependency. The repository installer renders `{{PAPER_SEARCH_DIR}}` to that dependency's absolute directory. Before running a command, check that the placeholder was rendered and the directory exists. If `{{PAPER_SEARCH_DIR}}` remains literal, stop and tell the user to rerun the repository installer. ## Command prefix ```bash PYTHONIOENCODING=utf-8 uv run --directory "{{PAPER_SEARCH_DIR}}" paper-search ``` Do not assume `paper-search` is globally installed. Use the full prefix so the pinned dependency environment is selected consistently. `PYTHONIOENCODING=utf-8` is required, not cosmetic. The CLI writes extracted paper text straight to stdout; on a Windows console the default cp1252 codec raises `UnicodeEncodeError` on the first mathematical or Greek character, and `read` returns `{"status": "error", ...}` instead of the paper. Any paper with an equation in it will trip this. On PowerShell use `$env:PYTHONIOENCODING = "utf-8"` before the call. ## Search ```bash uv run --directory "{{PAPER_SEARCH_DIR}}" paper-search search "<query>" \ --max-results <results-per-source> \ --sources <comma-separated-sources> \ [--year <year-or-range>] ``` - Default `--max-results` is 5 per source. - Default `--sources` is `all`. Prefer a focused set such as `arxiv,semantic,crossref,openalex` for an initial pass; broaden only when coverage matters more than latency. - `--year` filter