Anyura
Join usLog inTalk to us
AI Native

Clinic work, approved once.

A single Anyura AI action can find the right patients, prepare the operation, and wait for staff approval before anything is sent.

Native MCP

Anyura speaks fluent agent.

Drop the MCP URL into Claude, Cursor, ChatGPT, or your own agent. Every module exposes typed tools — bookings, charts, billing, recall sequences — with permission scopes you grant per-agent.

Live
›Front-desk · Recall draft
Agent reply
“Hi Hannah — it's been 12 weeks since your last visit. Same Friday slot work? Reply Y to confirm.”Queued for your review · no send until you tap approve
31 tools · 8 namespaces
MCP / JSON-RPC over SSE
anyura.calendar4
find_slotbookreschedulecancel
anyura.patients4
lookupcreateupdatetimeline
anyura.charts4
readappend_notephotoslock
anyura.billing4
chargerefundcalculatepost
anyura.inventory4
lotslookupadjustreorder
anyura.messaging4
draft_smssendthreadtemplates
anyura.marketing3
recall_seqattributionwinback
anyura.reports4
productionretentionno_showslot_variance
Read by default · write on grant
Agents see only what you allow. Write scopes (book, charge, send) require explicit per-agent approval — revoke in one click.
Every call signed + auditable
HIPAA-grade audit log. Each tool call carries agent identity, prompt context, and is reversible from the journal.
Typed schemas. Streaming.
Strict TypeScript schemas — the docs are the types. Long-running ops stream progress over SSE.