Safety and trust

This page sets out the current status of the project, the safety rules enforced by tests, and the use of AI in its development.

This system must not be used with real patients at present. The path to production explains what production means, with an estimate for each thing still needed.

Current status

The software is built and tested. The work queue has 52 open tasks. The main outstanding items require people or outside parties:

Figures on this site come from the repository's generated status, last retrieved on November 17, 2026.

Rules enforced by tests

Support only

No score may reduce what a patient is offered, reorder or withhold a slot, overbook, or delay a contact.

Contact after a missed appointment

A missed appointment triggers contact, never discharge. Urgent-pathway services hold a cancellation for a person to review.

Consent

Opt-outs and quiet hours are enforced at the one point every message passes through.

Gated features stay off

A gated feature does nothing until a switch is set and named people have approved it.

Safe reply links

Links are single-use, with hashed tokens. Opening a link reads data; only a form post changes anything.

How AI was used

Much of the repository was written with an AI assistant, Anthropic's Claude, at the maintainer's direction. That includes code, tests, specifications, governance drafts and translations.

Read the full AI statement.

Frequently asked questions

Is this an official health service product?

No. It is an independent open-source project, run by no health service or government. It began with the needs of hospitals in Wales and is being made country-neutral, so that one country's rules are a setting, not the core. This is not the website of any health service.

Can my organization use it?

The licenses permit it, but the system is not ready for real patients. See the path to production.

Does it store medical records?

No. It is administrative only.

Governance documents

These are drafts and self-assessments, not sign-offs.