Skip to content

Repository files navigation

uni-course-scheduler

AI-powered university course planning assistant — analyze course catalogs, get smart course recommendations, build conflict-free weekly timetables, and export Excel (local + cloud) and ICS calendar files (cloud standard mode only).

License

Language / 语言: English | 中文


What it does

uni-course-scheduler is an AI Skill that helps university students plan their semester:

  • Catalog analysis — parses a school's official course catalog (program handbook text) into structured course data: codes, names, credits/points, prerequisites, term availability, and timetable slots.
  • Course recommendation — recommends courses matched to the student's goals (e.g. GPA-first, ML specialization), flags enrollment priority, and validates prerequisites.
  • Weekly schedule — builds a conflict-free weekly timetable from the real timetable, respecting preferences like "no 8am classes" or "keep 12:00–13:00 free".
  • Excel export — generates a 6-sheet (local) or 7-sheet (cloud) Excel workbook, including Weekly Timetable, Degree Planner, and (cloud) Degree Audit & Risk Assessment.
  • ICS calendar — exports a downloadable .ics file for Apple Calendar / Google Calendar (cloud mode).

Supports international credit systems (US / AU / UK / ECTS / CN / HK / SG) and undergraduate and graduate students.


How it works (two modes)

This Skill ships in two execution modes:

1. Local personal mode (free)

The agent itself researches the official course catalog, verifies it, and produces a 6-sheet Excel + a textual weekly overview.

  • Free, single-school, single-student, personal use only.
  • No batch pipeline, no standardized decision report, no downloadable .ics calendar, no deep degree audit.

2. Cloud standard mode (paid)

A fixed production-grade cloud pipeline (catalog → recommend → schedule → decision → ICS → degree-audit) processes the catalog text and returns:

  • Batch / multi-school / multi-student execution.
  • Standardized 8-dimension Decision Report.
  • Degree Audit & Risk Assessment sheet (prerequisite chains, credit progress, overload risks).
  • Downloadable .ics calendar (university term dates, weekly recurrence, correct IANA timezone).

Offline-cloud note: Like most hosted LLM pipelines, the cloud executor has no internet access. The local agent (or the user) must first paste the official catalog text. The pipeline never fabricates data — anything not found in the input is reported honestly as missing.


Cloud platform (choose your payment)

The cloud command is platform-agnostic; the platform is chosen by your payment method:

Your payment Platform Price Server Token env
China mainland payment (Alipay/WeChat/UnionPay) Shengsuanyun ¥0.5/run loomloom.shengsuanyun.com LOOMLOOM_TOKEN_SHENGSUANYUN
International credit card (Visa/MC) CogFoundry $0.10/run loomloom.cogfoundry.ai LOOMLOOM_TOKEN_COGFOUNDRY

On first cloud use, the assistant asks which payment you use and configures the matching server. See references/loomloom-setup.md for full setup.


Install

Prerequisites

  • An AI assistant that supports custom Skills (e.g. WorkBuddy / Claude Code / Codex / Cursor-style agent).
  • For cloud mode: a platform account + balance.

Option A — drag & drop (friendliest)

  1. Download the uni-course-scheduler.zip release.
  2. Drag the ZIP into your AI chat and say: "Please install this planning Skill for me."
  3. The AI agent installs it and you can start.

Option B — manual install

# 1. Unzip the package
mkdir -p ~/.workbuddy/skills && cd ~/.workbuddy/skills
unzip /path/to/uni-course-scheduler.zip

# 2. Verify the folder structure
ls ~/.workbuddy/skills/uni-course-scheduler   # SKILL.md, scripts/, references/

Cloud setup (optional, for cloud mode)

# Install the LoomLoom CLI (buyer flow)
# See references/loomloom-setup.md for full instructions
loomloom login          # sign in with your key
loomloom market list    # verify connectivity

Usage

Plan a course schedule

  1. Tell the agent your school, major, year, and goals (e.g. "GPA first, prefer no 8am classes").
  2. Provide the official catalog text — let the local agent fetch it, or paste it.
  3. Choose a mode: Local (free) or Cloud (paid).
  4. Confirm the fee (cloud) and receive: an Excel workbook (.xlsx) + a downloadable .ics calendar (cloud).

Collect a course catalog

See references/catalog-collection.md for a reusable prompt and quality gates (≥500 chars, ≥5 courses, official source).


Project structure

uni-course-scheduler/
├── SKILL.md                      # Main skill instructions (both modes)
├── README.md                     # This file (English)
├── README.zh-CN.md               # Chinese version of this README
├── LICENSE                       # Personal-use license
├── agents/
│   └── openai.yaml               # OpenAI / compatible agent metadata
├── references/
│   ├── catalog-collection.md     # How to collect & verify catalogs
│   ├── excel-output-spec.md      # Excel workbook schema
│   ├── cloud-output-format.md    # Cloud pipeline output format
│   ├── interaction-flow.md       # End-to-end user flow
│   ├── local-validation.md       # Local output validation rules
│   └── loomloom-setup.md         # Cloud setup (dual-platform)
└── scripts/
    ├── generate_excel.py         # Build the Excel workbook from JSON
    ├── generate_ics.py           # Render .ics from cloud ICS data
    ├── anti_hallucination_check.py  # Anti-hallucination validation
    ├── validate_schedule.py      # Schedule conflict validation
    ├── local_audit.py            # Local output audit
    └── check_github_downloads.py # Download stats helper

Quality & anti-hallucination

  • No fabrication: data not present in the input is labeled NOT_FOUND / missing — never invented.
  • Honest empty: for garbage/empty input returns an honest empty result with guidance, not a hallucination.
  • Validation: local scripts run schedule-conflict checks and an anti-hallucination audit on every output.

License

Personal Use License — No Commercial Redistribution. Free for personal use; commercial redistribution or re-listing on paid marketplaces requires the author's written permission. See LICENSE (Chinese text prevails).


Support / Feedback

  • Author: ez-hq
  • Found a bug or have a catalog request? Open an issue in the repository.

About

AI-powered university course planning Skill: analyze course catalogs, recommend courses, build conflict-free weekly timetables, export Excel + ICS. Supports US/AU/UK/ECTS/CN credit systems.

Topics

Resources

Stars

1 star

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages