GitHub - rizalfanex/skripsiku

AI Academic Writing Assistant

Powered by Kimi K2 × NVIDIA — for Indonesian students, researchers & journal authors

🚀 Quick Start · ✨ Features · 🐛 Report Bug · 💡 Request Feature

Warning

This project is in Beta — Features may be incomplete, APIs may change, and rough edges exist. Feedback and contributions are warmly welcome!

What is Skripsiku?

Skripsiku is a full-stack AI assistant designed specifically for academic writing in Indonesian universities. It helps you write, revise, paraphrase, analyze, and improve academic documents — from undergraduate theses to international journal papers — all powered by NVIDIA's Kimi K2 model family.

The interface is inspired by ChatGPT: you open the app and you're immediately in a chat. No mandatory project setup, no friction. Just start writing.

Features

🧠 Three AI Modes

Mode	Model	Best For
Instant	Kimi K2 Instruct	Fast drafting, grammar, paraphrasing, translation
Thinking Standard	Kimi K2 Thinking	Research gap analysis, argument expansion, critiquing
Thinking Extended	Kimi K2 Thinking (multi-step)	Journal upgrades, reviewer simulation, deep methodology review

✍️ Academic Task Library (14+ tasks)

Grammar correction & academic rewriting
Paraphrasing & similarity reduction
Research gap analysis & novelty identification
Hypothesis & conceptual framework drafting
Literature review synthesis
Methodology drafting & results interpretation
Abstract generation
Journal upgrade (Q1/Q2 targeting)
Reviewer simulation & rebuttal letter writing
APA / IEEE / MLA / Chicago / Harvard citation formatting
ID ↔ EN translation
Originality analysis

💬 ChatGPT-like Conversation UX

Land directly in chat — no project creation required
Full conversation memory: previous messages are sent as context
AI-generated conversation titles
Sidebar history grouped by: Today, Yesterday, This Week, Older
Projects are optional tags (for organizing chats), not mandatory gateways

⚡ Technical Highlights

Streaming via Server-Sent Events (SSE) — words appear in real time
Guest mode — works without any account
Auth — JWT with refresh tokens (optional but available)
Persistent history — conversations stored in SQLite (or PostgreSQL for production)
Docker-first — single docker compose up --build to run everything
Bilingual — Indonesian (PUEBI/EYD) and English academic output

Tech Stack

Layer	Technology
Frontend	Next.js 14 (App Router), TypeScript, Tailwind CSS, Framer Motion, Zustand
Backend	FastAPI, SQLAlchemy (async), SQLite / PostgreSQL
AI	NVIDIA API — Kimi K2 Instruct + Kimi K2 Thinking
Auth	JWT (access + refresh tokens)
Deployment	Docker + Docker Compose

Getting Started

Prerequisites

Docker & Docker Compose
An NVIDIA API key with access to Kimi K2 models

1. Clone the repository

git clone https://github.com/rizalfanex/skripsiku.git
cd skripsiku

2. Configure environment variables

# Copy the example and fill in your values
cp .env.example backend/.env

Open backend/.env and set at minimum:

NVIDIA_API_KEY=your-nvidia-api-key-here
SECRET_KEY=your-random-secret-key   # generate: python -c "import secrets; print(secrets.token_hex(32))"

3. Run with Docker

docker compose up --build

That's it. The database is created automatically on first run.

Local Development (without Docker)

Backend

cd backend
python -m venv venv
# Windows: venv\Scripts\activate | Mac/Linux: source venv/bin/activate
pip install -r requirements.txt
cp ../.env.example .env   # then edit .env
uvicorn main:app --reload --host 0.0.0.0 --port 8000

Frontend

cd frontend
npm install
# Create frontend/.env.local:
echo "NEXT_PUBLIC_API_URL=http://localhost:8000" > .env.local
npm run dev

Project Structure

skripsiku/
├── backend/
│   ├── app/
│   │   ├── api/v1/          # FastAPI routers (auth, projects, chat, conversations, export)
│   │   ├── core/            # Config, database, security
│   │   ├── models/          # SQLAlchemy ORM models
│   │   ├── schemas/         # Pydantic request/response schemas
│   │   └── services/llm/    # Orchestrator + NVIDIA API client
│   ├── Dockerfile
│   └── main.py
├── frontend/
│   ├── src/
│   │   ├── app/             # Next.js App Router pages
│   │   │   ├── (auth)/      # Login, Register
│   │   │   └── (dashboard)/ # Chat, Projects, Settings, Workspace
│   │   ├── components/      # UI & layout components
│   │   ├── hooks/           # useChat, useConversations, useProject
│   │   ├── lib/             # API client, types, utilities
│   │   └── store/           # Zustand global state
│   └── Dockerfile
├── docker-compose.yml
└── .env.example

How It Works

AI Mode Routing

Each message goes through the Orchestrator, which routes to the appropriate model and pipeline:

Instant → single call to kimi-k2-instruct, streamed token by token
Thinking Standard → single call to kimi-k2-thinking with extended reasoning
Thinking Extended → multi-step pipeline: Initial Draft → Academic Reasoning → Final Revision

Conversation Memory

When authenticated, every user message and assistant response is persisted to the database. On revisiting a conversation, the full message history is loaded from the backend and sent as context in subsequent requests — exactly like ChatGPT.

Streaming

Responses stream via SSE. The frontend renders markdown in real-time using react-markdown. A special meta event carries the conversation_id so the URL updates immediately without a page reload.

Roadmap

File/PDF upload and extraction
Citation database search (Semantic Scholar, CrossRef)
Export to DOCX (in progress)
Collaborative workspaces
Fine-tuning on Indonesian academic corpus
Mobile-responsive UI improvements
Usage analytics dashboard

Contributing

This project is in early beta and welcomes contributions of all kinds:

Fork the repository
Create a feature branch: git checkout -b feature/your-feature
Commit your changes: git commit -m 'feat: add your feature'
Push to your branch: git push origin feature/your-feature
Open a Pull Request

Please open an issue first for major changes.

License

This project is licensed under the MIT License. See LICENSE for details.

Disclaimer

Skripsiku is a tool to assist academic writing, not to replace the writer. Always review, verify, and take responsibility for the content you submit. Academic integrity is your responsibility.

Made with ☕ for Indonesian academia

Arsitektur

skripsiku/
├── backend/               # FastAPI + SQLAlchemy
│   ├── app/
│   │   ├── api/v1/        # Auth, Projects, Chat (SSE), Export
│   │   ├── core/          # Config, Database, Security
│   │   ├── models/        # User, Project, Conversation, Message
│   │   ├── schemas/       # Pydantic request/response schemas
│   │   └── services/
│   │       ├── llm/       # NVIDIA provider + Orchestrator
│   │       └── prompts/   # Templates + Builder
│   └── main.py
│
└── frontend/              # Next.js 14 + TypeScript + Tailwind
    └── src/
        ├── app/
        │   ├── page.tsx              # Homepage
        │   ├── (auth)/               # Login, Register
        │   └── (dashboard)/          # Dashboard, Projects, Workspace, Settings
        ├── components/ui/            # Button, Input, Card, Badge, Modal, Spinner
        ├── hooks/                    # useChat, useProject
        ├── lib/                      # types, utils, api client
        └── store/                    # Zustand global store

Multi-Model Orchestration

Mode	Model	Pipeline
Instant	`kimi-k2-instruct`	Single call — drafting, grammar, paraphrase, translate
Thinking Standard	`kimi-k2-thinking`	Single call with analysis overlay — research gap, discussion
Thinking Extended	instruct → thinking → instruct	3-step pipeline — journal upgrade, thesis generation

Beberapa task type secara otomatis di-upgrade ke mode yang lebih tinggi terlepas dari pilihan pengguna (contoh: research_gap_analysis → Thinking Standard minimum).

Setup

Prasyarat

Python 3.11+
Node.js 18+
NVIDIA API Key (daftar di https://build.nvidia.com)

1. Clone & Environment

git clone <repo-url>
cd skripsiku
cp .env.example .env

Edit .env dan isi nilai berikut:

NVIDIA_API_KEY=nvapi-xxxxxxxxxxxxxxxxxxxx
SECRET_KEY=ganti-dengan-string-acak-panjang-minimal-32-karakter

2. Backend

cd backend
python -m venv venv

# Windows
venv\Scripts\activate
# Linux/Mac
source venv/bin/activate

pip install -r requirements.txt

uvicorn main:app --reload --host 0.0.0.0 --port 8000

Backend API tersedia di: http://localhost:8000
Dokumentasi interaktif: http://localhost:8000/docs

3. Frontend

cd frontend
npm install
npm run dev

Aplikasi tersedia di: http://localhost:3000

Konfigurasi `.env`

Variabel	Default	Keterangan
`NVIDIA_API_KEY`	—	Wajib. API key dari build.nvidia.com
`NVIDIA_BASE_URL`	`https://integrate.api.nvidia.com/v1`	Endpoint NVIDIA API
`NVIDIA_DEPLOYMENT_MODE`	`hosted`	`hosted` atau `self-hosted` (NIM)
`MODEL_INSTANT`	`moonshotai/kimi-k2-instruct-0905`	Model untuk mode Instant
`MODEL_THINKING_STANDARD`	`moonshotai/kimi-k2-thinking`	Model untuk Thinking Standard
`MODEL_THINKING_EXTENDED`	`moonshotai/kimi-k2-thinking`	Model untuk Thinking Extended
`DATABASE_URL`	`sqlite+aiosqlite:///./skripsiku.db`	Dapat diganti ke PostgreSQL
`SECRET_KEY`	—	Wajib. Kunci JWT (min. 32 karakter)
`BACKEND_CORS_ORIGINS`	`http://localhost:3000`	URL frontend (pisahkan dengan koma)
`MAX_TOKENS_INSTANT`	`4096`	Batas token mode Instant
`MAX_TOKENS_THINKING_STANDARD`	`8192`	Batas token Thinking Standard
`MAX_TOKENS_THINKING_EXTENDED`	`16384`	Batas token Thinking Extended

Docker (Opsional)

cp .env.example .env
# Edit .env dengan API key Anda
docker-compose up --build

Task Types yang Tersedia

Task	Deskripsi
`academic_rewrite`	Tulis ulang teks secara akademik
`grammar_correction`	Koreksi tata bahasa
`paraphrasing`	Parafrase dengan mempertahankan makna
`reduce_similarity`	Kurangi kemiripan tanpa mengubah substansi
`expand_argument`	Kembangkan argumen ilmiah
`thesis_title_generation`	Generate judul penelitian
`problem_formulation`	Susun rumusan masalah
`research_gap_analysis`	Analisis celah penelitian
`literature_review_synthesis`	Sintesis tinjauan pustaka
`methodology_drafting`	Draft metodologi penelitian
`discussion_strengthening`	Perkuat bab diskusi
`abstract_generation`	Buat abstrak terstruktur
`conclusion_improvement`	Perbaiki bab kesimpulan
`journal_upgrade`	Tingkatkan ke standar jurnal internasional
`reviewer_simulation`	Simulasi komentar reviewer
`cover_letter`	Tulis cover letter pengiriman jurnal
`rebuttal_letter`	Balas reviewers secara profesional
`translate_id_to_en`	Terjemahkan ID → EN akademik
`translate_en_to_id`	Terjemahkan EN → ID formal
`citation_formatting`	Format daftar referensi sesuai gaya
`originality_analysis`	Analisis orisinalitas argumen

Catatan Etika

Skripsiku dirancang untuk meningkatkan kualitas tulisan, bukan untuk:

Menghindari deteksi plagiarisme
Menghasilkan karya yang sepenuhnya bukan milik pengguna
Memfabrikasi referensi akademik

Semua prompt AI secara eksplisit mengedepankan integritas akademik.

Stack Teknologi

Backend: FastAPI · SQLAlchemy 2.0 · SQLite/PostgreSQL · JWT · httpx · tenacity · python-docx
Frontend: Next.js 14 · TypeScript · Tailwind CSS · Zustand · Framer Motion · Radix UI · react-markdown
AI: NVIDIA API · Kimi K2 Instruct · Kimi K2 Thinking

Name		Name	Last commit message	Last commit date
backend		backend
frontend		frontend
.env.example		.env.example
.gitignore		.gitignore
CHANGELOG.md		CHANGELOG.md
LICENSE		LICENSE
README.md		README.md
docker-compose.yml		docker-compose.yml
logo-skripsiku.png		logo-skripsiku.png
logo.png		logo.png

Latest commit

History

Folders and files

Repository files navigation