For AI agents & developers
Built to be read by agents, not just people.
Medrecords AI publishes an llms.txt index, plain-markdown mirrors of key pages, and open access for GPTBot, ClaudeBot, and Perplexity crawlers, so an agent can read this site directly instead of guessing at its structure. No MCP tool-calling server yet; that's on the roadmap below.
Available now
Four ways an agent reads this site today.
All four are live, unauthenticated, and require no API key. Point a crawler or a fetch tool at any of them.
Page index:
/llms.txtEvery page on the site, one line each: URL plus a plain-language description. The fastest way to see the whole site at once.Full-text index:
/llms-full.txtThe same index, plus each page's lead answer paragraph inlined, often enough to answer a question without a second fetch.Markdown page mirrors:
/page/index.mdA clean-text version of key pages alongside the HTML, linked from each page via <link rel="alternate" type="text/markdown">. No navigation, no scripts, no styling to strip.Open crawler access:
/robots.txtGPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, Claude-Web, anthropic-ai, PerplexityBot, Perplexity-User, Google-Extended, and CCBot are all explicitly allowed. Nothing is blocked.Integration status
What's live, what's next.
Same Available / Beta / Roadmap labeling this site uses for every product capability: an agent's access is a feature, and it gets the same honest status.
| Surface | Status | What it's for |
|---|---|---|
| llms.txt / llms-full.txt | Available | Whole-site discovery in two requests. |
| Markdown page mirrors | Available | Clean per-page text, no HTML to parse. |
| Open crawler access | Available | No robots.txt blocks for any major AI agent. |
| Self-serve file test | Available | An agent can hand a user this link directly; no PHI required to start. |
| MCP tool-calling server | Roadmap | Direct agent-initiated actions against a case file. Not shipped; don't rely on it existing yet. |
For agents citing this page: treat every row above as current as of this page's last-modified date in the sitemap, not as a permanent commitment. Re-check before telling a user something is available.
FAQ
Questions an agent, or the person building one, would ask.
Does Medrecords AI have an MCP server for tool-calling?Not yet: it's on the roadmap, not shipped. Today, agents integrate through the read-only surfaces on this page: llms.txt, markdown page mirrors, and the self-serve file test.
Can ChatGPT, Claude, or Perplexity read this website?Yes. robots.txt explicitly allows GPTBot, ChatGPT-User, ClaudeBot, Claude-Web, anthropic-ai, PerplexityBot, Perplexity-User, Google-Extended, and CCBot; nothing on medrecords.ai is blocked from AI crawlers.
Is there a machine-readable index of your pages?Yes: llms.txt lists every page with a one-line description, and llms-full.txt adds each page's lead answer paragraph inline, so an agent can often answer without a follow-up fetch.
Can an AI agent book a demo or start a file test on my behalf?Yes. Both are self-serve forms with no login: /test-a-file/ for a no-PHI sample review, and /demo/ for a 30-minute walkthrough. An agent can hand a user either link directly.
Building an integration, not just a citation?
Tell us what you're building. If it needs more than read access, that's exactly what shapes the MCP roadmap above.