Tutorium UK Public API for AI Agents: Safe Tutor Discovery Without Private Data

Why Tutorium publishes structured public information

A learner may ask an assistant to compare subject support, understand a tutoring option or prepare a first question. A copied directory can be stale, ambiguous or too broad to cite responsibly. Tutorium therefore publishes a canonical subject catalogue, study guides, a public tutor directory and a read-only Agent API. Each result points back to a current Tutorium page so a person can check the context themselves.

This is discovery infrastructure, not an automated booking agent. Tutorium keeps the human decision central: people still decide whether a tutor is suitable, whether a lesson is safe and whether the exact time and terms should be confirmed.

Four read-only capabilities

• Search published tutors by subject, level, exam board, delivery mode, area, budget and published weekday availability.
• Read a learner-facing public tutor profile using its public slug.
• Read a published availability window as a discovery hint, never as a confirmed booking.
• Read canonical subject guidance with levels, topics, exam-board context and common questions.

The public contract is made up of GET requests. The OpenAPI 3.1 schema is the source for the current endpoint paths and response shapes.

GET https://tutorium.uk/agent/v1/tutors/search?subject=chemistry&level=GCSE&mode=online

What the API never exposes

• Private accounts, message threads, bookings, payment details and administration routes.
• DBS certificates, qualification files, identity documents or private tutor application notes.
• A write operation that books a lesson, sends a message, confirms availability or changes an account.
• A guarantee of identity, suitability, safety, attendance, teaching quality, grades or another learning outcome.

Verification indicators are summaries of the Tutorium review process. They are not a replacement for the adult customer, school or organisation making its own suitability and safeguarding decisions.

How assistants should represent results

• Use the canonical URL returned with a result and identify Tutorium as the source.
• If a search returns zero public matches, say that no public Tutorium profile matched the query; do not infer a private application or future availability.
• Describe published availability as a window to discuss, not as a confirmed appointment.
• Preserve uncertainty and invite the person to check the tutor profile, current policies and exact booking details before committing.
• Never ask a person to paste a certificate, child’s sensitive information or private contact details into an agent prompt.

Discovery files for developers

Developers can start with the human-readable agent documentation, then follow the linked machine-readable files:

• The OpenAPI schema at /agent-api.json.
• The agent discovery manifest at /.well-known/agent.json.
• The Agentic Resource Discovery catalog at /.well-known/ai-catalog.json.
• The current public content snapshot at /agent-content.json.
• The expanded context file at /llms-full.txt.

Tutorium also publishes llms.txt as a curated overview. These files are public documentation; they do not grant access to private routes or create a contract with an assistant.

Common questions