SoCSE-Curriculum

Curriculum-design system for REVA University, School of Computer Science & Engineering (SoCSE) — the 2026 B.Tech scheme (CSE, ISE, IT, AIML, AIDS, IoT, Cyber Security).

This repository is not software — it is a knowledge base + AI workflow that helps faculty design and review dual-level Course Files with the help of an AI coding assistant (Claude Code, VS Code, or Antigravity IDE).

The one idea this repo serves — dual-level design. Every core course is delivered at two levels in one course: an awareness floor every student must clear (protecting placement readiness) and an advanced ceiling needed to exceed 8 CGPA — separated by assessment, not by streaming students into sections. Read REVA_BTech_Curriculum_Strategy.md for the why.

---

What's in here

Document	What it is
REVA_BTech_Curriculum_Strategy.md	The strategy — the why behind dual-level design
Course_Designer_Guidelines_2026.md	Design rules by course category and L-T-P pattern
Course_File_Template.md	The Course File template faculty fill (§0–§22)
Curriculum_Visual_Map.md	Semester map: each course's category, L-T-P-C, level, prerequisites
Course_Design_Workflow.md	The two-path faculty workflow (start from Outcomes or Activities)
Course_Design_Checklist.md	The ~45-item gate every finished course must pass
Course_Design_Verification_DAA.md	A worked example: the DAA course file verified against the checklist
.claude/skills/	Two AI skills: reva-course-designer and reva-course-reviewer
AGENTS.md	Orientation for AI agents working in this repo
CONTRIBUTING.md	How to contribute (setup, GitHub, architecture)

---

Before you start

1. A GitHub account (a GitHub-for-Education account is ideal).
2. Git installed — https://git-scm.com/downloads.
3. One AI assistant of your choice (set up below): Claude Code, VS Code + an AI extension, or Antigravity IDE.
4. (Optional) Python 3 — only needed if a tool must read existing .docx course files.

Clone the repository:

git clone https://github.com/kavyareva24/SoCSE-Curriculum.git
cd SoCSE-Curriculum

---

How to design a course (pick your tool)

All three tools do the same job — open this folder so the AI can read the strategy, template, guidelines, and map, then drive the two-path workflow. The difference is only how the AI is invoked.

Option A — Claude Code (recommended; skills work natively)

Claude Code auto-discovers the skills in .claude/skills/.

cd SoCSE-Curriculum
claude

Then just describe your task in plain English — the right skill triggers automatically:

• To design: "Design the course file for Operating Systems (Sem 5). I'll give you the course outcomes." → triggers reva-course-designer.
• Activities-first: "Here are the weekly activities for my course — reverse-engineer the COs and the full course file."
• To review: "Verify my DAA course file against the checklist." → triggers reva-course-reviewer.

You can also type / to see available skills. Claude reads AGENTS.md for repo context.

Option B — VS Code

1. Install VS Code, then add an AI assistant extension — the Claude Code extension (skills work the same as Option A) or GitHub Copilot Chat.
2. File → Open Folder… → select the cloned SoCSE-Curriculum folder.
3. Open the AI chat panel. With the folder open the assistant can read the template, guidelines, strategy, and AGENTS.md.
4. Drive the workflow by pasting a prompt from Course_Design_Workflow.md §2 (Outcomes-first) or §3 (Activities-first) — e.g.:

   "Using Course_File_Template.md, Course_Designer_Guidelines_2026.md, and the curriculum strategy, generate a complete Course File for ‹code – title› (category ‹…›, L-T-P ‹…›, level ‹…›). My COs and Bloom levels are below…"

5. Attach your existing/old course document if you have one, and ask the AI to modernise it onto the template.

With the Claude Code extension, the reva-course-designer / reva-course-reviewer skills trigger just like in Option A. With Copilot, paste the workflow prompts manually instead.

Option C — Antigravity IDE

Antigravity is an agent-first IDE.

1. Open Antigravity and open the cloned SoCSE-Curriculum folder as your workspace.
2. Open the Agent panel. Antigravity reads AGENTS.md for repo context automatically.
3. The .claude/skills/ files don't auto-trigger here (they're a Claude Code convention), so drive the process with the prompts in Course_Design_Workflow.md — point the agent at the template, guidelines, strategy, and map, and follow Path A or Path B.
4. Let the agent fill Course_File_Template.md interactively, then review.

---

The workflow in one picture

Locate your course in the map (category, L-T-P, level, prerequisites)
                              │
              ┌───────────────┴───────────────┐
        PATH A: write Course Outcomes    PATH B: list Activities in the
        (conventional OBE)               session plan (one per session)
              └───────────────┬───────────────┘
                              ▼
        AI fills the dual-level Course File (interactive or one-shot)
                              ▼
        FACULTY review (human first)  →  AI review (best practice +
                              ▼           strategy + curriculum structure)
        Verify against Course_Design_Checklist.md  →  submit to BoS

Full detail: Course_Design_Workflow.md.

---

Reviewing / verifying a finished course

Ask your AI assistant to run the course through Course_Design_Checklist.md (the reva-course-reviewer skill does this in Claude Code). It produces a per-item Pass / Fail verdict with fixes and an overall READY / NOT READY result. See Course_Design_Verification_DAA.md for the format and rigour expected.

---

Contributing & ownership

The curriculum is owned by REVA SoCSE. To propose changes — a new course file, a template/guideline update, or a new skill — read CONTRIBUTING.md for setup, the GitHub pull-request process, and the repository architecture.