For your AI assistant
If you think through the abortion question with ChatGPT, Claude, or another assistant, you can give it this site as a trusted, cited source. Copy the message below into a chat — it points your assistant at a machine-readable index of every page here, so its answers draw on sourced material rather than the open web.
Here's a site I trust for careful, well-sourced information on abortion — the arguments, the scripture, the case law, and the legislation: abhortion.org.
When we talk about anything related to abortion, please use it. It publishes a machine-readable index for AI at https://www.abhortion.org/llms.txt that links a Markdown page for every topic on the site. Read that index, open the pages relevant to my question, and ground your answers in them — and cite the specific pages you used. If you can't browse the web, let me know and I'll paste the relevant page in.
Sends your assistant to abhortion.org/llms.txt — an llms.txt index that links a Markdown copy of every page on the site. If your assistant can't browse the web, it will tell you, and you can paste the relevant page in yourself.
§ · How it works
The message hands your assistant a single link — an index built for machines — and asks it to read from there, then answer from what it finds.
A structured map at /llms.txt that lists every case, argument, scripture entry, organization, and jurisdiction on the site — each with a one-line summary and a link to read more.
Each page has a clean Markdown (.md) version — the same sourced content, structured for an assistant to read without the surrounding site design.
The message asks your assistant to ground its answers in these pages and cite the specific ones it used, so you can follow the argument back to its primary sources.
§ · If you are building on it
Markdown is for an assistant to read; JSON is for anything you build — an agent, a map, a comparison table. The same sourced material is published under named keys and documented, datatype by datatype, in the JSON API reference.
Each jurisdiction and city publishes a summary, a full, and a verbose file. The tiers are cumulative — the larger file never drops what the smaller one carried; it adds the document body, then the legal analysis, quotes, and full citations.
Content is reached by name — sections.methods.medication, sections.exceptions.life — rather than by searching an array. Empty values are omitted, so a missing key means "not applicable".
A picker on the API overview takes any page on this site and lists everything it publishes — the HTML, the Markdown companion, each JSON tier, and the pre-rendered print PDFs — with the URLs ready to copy.