ANU Semester Planner

A small full-stack replacement for one frustrating part of ANU course planning: today a student has to look up course codes, course pages, semester availability and study-load information from several separate places just to sketch out a semester. This prototype puts a small, representative catalogue of computing courses in one place and lets a student build and save a semester plan against it.

A student types in a Student ID, which gets or creates their plan — no password, and Student ID alone is not authentication, just a simple identifier for this prototype. From there they can add courses to Semester 1 or Semester 2 from a catalogue of the 2026 Master of Computing (Software Development) courses, see live per-semester unit totals, and remove courses they no longer want. Everything is written to SQLite as it happens, so refreshing the page or coming back later with the same Student ID always shows the same plan.

The catalogue also carries real ANU prerequisite and incompatibility rules for these courses, sourced from each course’s own Programs & Courses page (not the aggregated program-requirements page, which conflates COMP7710 and COMP6710). A course can’t be added until its enforceable prerequisites are met, either by an earlier semester in the same plan or by marking it “previously completed.” This is still a small demonstration catalogue, not the complete ANU course list or an authoritative source of enrolment rules.

What good looks like here

Good, for this prototype, means the full loop actually holds under a real reload: Student ID → plan loads or is created → a course is added → it’s in SQLite → the page reloads → the course is still there — and the same is true returning later with the same Student ID. That contract is what most of the spec tests in spec/ check, and it’s also what I re-verified by hand against the built server, not just trusted to the automated suite.

Decisions I made, and why:

• Real course data, not invented data. CLAUDE.md asks for a “representative” catalogue, but representative still means real, so the ten courses, their units, semester availability, and requisite text are all cross-checked against ANU’s own course pages rather than made up.
• Prerequisites are modelled as OR-within-a-group, AND-across-groups, because real ANU requisites aren’t a single chain — some courses need one of several alternatives, others need several independent things at once. Where a real requirement can’t be enforced with this schema (for example, a requirement that’s actually about program enrolment, or references a course outside this catalogue), the verbatim requisite text is still shown so the app never silently implies a course is prerequisite-free.
• A lightweight “previously completed” flag, short of any real academic history feature, exists only because some prerequisite courses in this catalogue are Semester-2-only, so an earlier-semester requirement can’t always be satisfied purely by in-plan sequencing within one modelled year.
• What I deliberately left out: ANU authentication/SSO, the complete ANU catalogue, complex/complete prerequisite enforcement, timetable clash detection, real enrolment submission, and anything to do with fees or academic records. These are out of scope for a prototype demonstrating one slice of the planning flow, per CLAUDE.md.

The rules behind these choices live in CLAUDE.md; the checks that hold the app to them live in spec/; the fuller account of how and why each decision was made, including mistakes caught along the way, is in PROCESS.md.