Methodology
How listings are chosen, what the regional evidence labels mean, which dates are recorded, and how to correct or remove a listing. The date added records when a listing entered the directory. Published records when a paper appeared. Pushed records when code last changed.
What gets listed
A publication needs a persistent DOI, PubMed ID or preprint identifier, a health or life-science subject, and a regional institution named in the affiliation metadata. Corrections and retraction notices are excluded. An eligible paper that has been retracted carries a retraction label. Linked preprint and journal versions have separate records and count once in publication totals.
A repository must contain its own work and have a plausible health, biomedical or life sciences purpose described in the repository or a linked source. A roster, an institution’s page or a releasing paper with a regional affiliation must connect the project to the region.
Empty repositories, coursework, personal configuration, unmodified forks and general-purpose mirrors are excluded.
Projects and publications use eight health topics. A publication that the topic rules cannot place in a health topic carries the label Other science. It remains searchable and appears on institution and county pages, but it has no topic page or topic feed.
The clinical studies index includes interventional trials, observational studies and expanded access records from ClinicalTrials.gov. A study needs at least one registry-listed facility in a supported county. We use the reported city and state, a corroborated institution site or a Census ZIP Code Tabulation Area contained in one county. Ambiguous locations are held until their county can be supported. Overall and site statuses come from the registry. Study listings can be searched by registry condition.
Regional evidence
For scholarly work, a Verified affiliation names the institution and a regional place, or a single-site regional institution. Listed indicates an indexer assignment or an institution match with an uncertain site. Company affiliations require a stated regional site. A county requires evidence placing the affiliation there. On multi-institution papers, filters match the institutions and counties recorded for the work. Follow the affiliation source to identify the site. A byline reports names as published.
- Verified
- A source ties the exact repository to the region.
- Listed
- A source shows a regional association through the GitHub account or lab.
A label describes the evidence for the regional connection. Follow the project’s regional source for the specific claim. Placement in the directory does not establish ownership, employment, a worksite or clinical validation.
Regional and repository coverage
Pennsylvania Bucks, Chester, Delaware, Montgomery, Philadelphia.
New Jersey Burlington, Camden, Gloucester, Mercer.
County counts include listings with a source-supported county. Activity uses the last code push when captured, or the GitHub repository update date otherwise. Repository update dates can reflect changes besides code. Missing language, license, paper and county fields are omitted from project summaries.
Corrections, removals and submissions
Email the directory team with the listing URL or persistent identifier and, for a correction or submission, a source for the regional connection. For a removal request, include the listing URL or directory ID and the reason for the request.
Citing a listing
Cite the publication or repository itself, using the publisher link or DOI on its page. To cite the regional connection, give the directory page URL, its evidence label and the date you accessed it. Counts on institution, topic and county pages are computed when the page is requested and change as listings are added or removed.
Machine-readable directory
Subscribe to directory additions. Institution, topic and county pages offer scoped feeds.
Download the project catalog as JSON. It returns a projects array. Use /api/directory/projects/{id} for a single project by GitHub repository ID.
Browse publication records as JSON. Results include distinct work and record totals, with up to 24 records per page.
Browse clinical studies as JSON. Look up one study with /api/directory/studies/{NCTId}.
Read directory additions as JSON. Use the institution, topic, county and kind filters to narrow a feed or JSON listing.
The endpoints support GET and HEAD. Removed projects are excluded from the catalog and return HTTP 410 at their own URL. The endpoints return HTTP 503 when removal records are unavailable.
Search engines and AI assistants may crawl and cite the public pages. robots.txt states the crawler policy, llms.txt summarizes the site for language-model tools, and the sitemap lists every public page.
Connect an AI assistant to search the directory through its read-only MCP server at /mcp.