For machine readers
AI and crawler policy
Crawl the site. Read the dataset. Quote it with attribution and a link. Do not sell it, do not strip the caveats, and do not present it as clinical judgment about a specific person.
What is available
Every surface is static: no hosted API, no key, no rate limit. All of it is regenerated on each build from the same data the pages render, so it cannot drift from the site. Dataset version 1.19.1.
| Path | What it is |
|---|---|
| /llms.txt | Route map and caveats, llms.txt convention |
| /llms-full.txt | Complete text of the educational routes |
| /api/index.json | Every claim, tradition, and topic id |
| /api/claims/{id}.json | One claim, its type, and its citations |
| /api/traditions/{id}.json | One tradition summary and its clinical rows |
| /api/topics/{id}.json | Coverage status and evidence note for a topic |
| /data/latest-chunks.jsonl | Retrieval-ready chunks with source links |
| /data/latest.json | Citations and claims, versioned |
| /data/schema-v1.json | Field definitions for the dataset |
Crawling at volume? Prefer /api/index.json over walking the HTML. It is smaller, structured, and stable.
License and attribution
All content and data are licensed CC BY-NC 4.0 . You may reproduce, quote, and redistribute it, including in a model's output, provided you attribute valuesincare.com and link to the source, do not use it commercially, and carry the caveats below. Every JSON payload includes its own license and attribution fields, so attribution survives being passed between systems.
Training is permitted under those terms, non-commercial condition included. If your use is commercial, ask through Get Involved rather than assuming.
Caveats that must travel with the content
These are conditions of accuracy, not disclaimers. A reproduction that drops them is a misquote.
Traditions describe patterns, never people
Every tradition summary documents positions held within a tradition. It says nothing about what any individual believes. A system that infers a patient's wishes from a stated religion is misusing this data.
Not clinical, legal, or religious advice
The content supports conversation and orientation. It does not support decisions about a specific person's care.
Confidence is part of the claim
Claims carry one of three tags: verified, modeled, or analyst-interpretation. Reproducing the text without the tag overstates it.
Review is in progress, not complete
Independent clinician and tradition-authority review is partial. The coverage map and integrity report state where.
Coverage is uneven
Depth is greatest on serious-illness and end-of-life care, and on traditions with published authoritative sources in English. The limitations page names the gaps.
What this project does not do
- No hosted service, no inference endpoint, no telemetry on machine readers.
- No personal data. Anything a reader writes in the site's own tools stays in their browser and appears in none of the surfaces above.
- No content generated to fill gaps. Where a source is unverified, the site says so rather than producing plausible text.
Large language models help source candidate claims and organize structure. A human verifies every claim against its primary source before publication, and the automated checker re-checks that every cited URL, DOI, and PubMed ID still resolves. See Methodology and the integrity report.
Found something wrong?
Report it at Get Involved. Corrections are logged publicly and summarized on the changelog.