$npx -y skills add ai4protein/VenusFactory2 --skill biorxivbioRxiv & medRxiv — biology / medicine preprint servers. Search by keyword + date window, fetch a specific preprint's full metadata (with all version history + abstract + JATS XML link) by DOI. Use for cutting-edge biology/medicine work that hasn't gone through peer review yet, o
| 1 | # bioRxiv & medRxiv |
| 2 | |
| 3 | ## Overview |
| 4 | |
| 5 | Two tools: existing `query_biorxiv_tool` (keyword/category date-window search) + new `download_biorxiv_by_doi` (fetch one preprint by DOI with all metadata, including version history). |
| 6 | |
| 7 | ## Project Tools (VenusFactory2) |
| 8 | |
| 9 | | Tool | Args | Returns | Description | |
| 10 | |------|------|---------|--------------| |
| 11 | | **query_biorxiv** | `query` (category or keyword), `max_results` (default 5, max 50), `days` (default 30; date window) | JSON list of paper records inline | Browse recent preprints in a category. | |
| 12 | | **download_biorxiv_by_doi** | `doi` (bare `10.1101/...` or DOI URL), `out_dir`, `server` (`biorxiv` \| `medrxiv`; default `biorxiv`), `include_abstract` (default True), `timeout` (default 30s) | rich JSON envelope; full metadata + version list at `file_info.file_path`; `biological_metadata` extracts title/authors/date/category/license/jatsxml URL | Fetch one preprint by DOI. | |
| 13 | |
| 14 | ## When to Use Each |
| 15 | |
| 16 | | Goal | Tool | |
| 17 | |---|---| |
| 18 | | "Latest bioinformatics preprints" | `query_biorxiv(query="bioinformatics", days=7)` | |
| 19 | | "Get me preprint X by DOI" | `download_biorxiv_by_doi` | |
| 20 | | Resolve a published paper back to its preprint version | `download_biorxiv_by_doi(doi=<published-DOI>)` (if a preprint exists with that DOI) | |
| 21 | | Get JATS XML for ML training data | `download_biorxiv_by_doi`, then GET the `jatsxml` URL from the metadata | |
| 22 | |
| 23 | ## Server Choice |
| 24 | |
| 25 | - **biorxiv** (default) — biology preprints (>200K papers, since 2013) |
| 26 | - **medrxiv** — medicine / health-sciences preprints (>40K papers, since 2019) |
| 27 | |
| 28 | If a DOI isn't found on biorxiv, try medrxiv (they share `10.1101/` DOI prefix). |
| 29 | |
| 30 | ## Version Handling |
| 31 | |
| 32 | `download_biorxiv_by_doi` returns *all* versions in `latest_versions[*]`. The top-level `latest` field is the most recent version (newest `version` number). Use `latest.jatsxml` for full-text XML, `latest.date` for publication date. |
| 33 | |
| 34 | ## Common Mistakes |
| 35 | |
| 36 | - **Wrong server**: if you get `NotFound`, try `server="medrxiv"` (or vice versa). |
| 37 | - **DOI URL not normalized**: the tool strips `https://doi.org/`, `http://doi.org/`, `doi:` prefixes — fine to paste a URL. |
| 38 | - **Trying to download the PDF directly**: this tool returns the *metadata JSON* (which includes a PDF URL). To download the PDF, follow the URL via `WebFetch` or `requests`. |
| 39 | - **Old DOI on a withdrawn preprint**: the API may still return metadata with `type="withdrawn"`. Check `category` / `type` fields if integrity matters. |
| 40 | |
| 41 | ## References |
| 42 | |
| 43 | - [bioRxiv API docs](https://api.biorxiv.org/) |
| 44 | - [bioRxiv home](https://www.biorxiv.org/) / [medRxiv home](https://www.medrxiv.org/) |