🌸 Hikari — Lightweight Coding Chatbot

Hikari is a free and open-source coding assistant built for people with older or lower-end computers.
It helps you write, fix, and explain code using modern open models that run locally — no internet required once installed.

This project was created to make AI-powered development accessible everywhere 🌍.

---

🧠 Features

• 💻 Runs entirely offline using Ollama
• ⚡ Works fast even on low-VRAM GPUs or CPUs
• 🪶 Simple, minimalist Japanese-inspired web UI
• 🧩 Supports small coding models like:
  • phi3:mini (Microsoft — very efficient)
  • codellama:7b-instruct
  • deepseek-coder:1.3b-instruct

---

🧰 Requirements

Before starting, make sure you have:

Tool	Description	Download
Python 3.10+	Needed for the backend (FastAPI)	python.org/downloads
Node.js 18+	Needed for the frontend build	nodejs.org
Ollama	Runs the AI model locally	ollama.com/download

🖥️ Recommended minimum hardware:
Ryzen 5 / i5 CPU, 8 GB RAM, optional GPU (4 GB VRAM or higher).

---

🚀 Quick Start

1️⃣ Clone or download Hikari

git clone https://github.com/YOURUSERNAME/hikari.git
cd hikari

---

2️⃣ Install the backend (AI server)

cd backend
python -m venv venv
venv\Scripts\activate
pip install -r requirements.txt

Then start the server:

python -m uvicorn main:app --host 0.0.0.0 --port 8080

If it works, you’ll see:

✓ Ollama is available
Uvicorn running on http://0.0.0.0:8080

---

3️⃣ Install Ollama and pull a model

Go to https://ollama.com/download and install it.

Then open a terminal and type:

ollama pull phi3:mini

🟢 Phi-3 Mini is small, fast, and great for writing code.

---

4️⃣ Build the web UI (frontend)

cd ../frontend
npm install
npm run build

When it finishes, you’ll see a new folder called dist/.

---

5️⃣ Open the chatbot

Go back to the backend folder and start Hikari again:

cd ../backend
python -m uvicorn main:app --host 0.0.0.0 --port 8080

Now open your browser and visit:

👉 http://localhost:8080

You should see your Hikari chat window appear 🌸
Type something like:

write me a hello world in python

---

🧩 Common Problems & Fixes

Problem	Cause	Fix
❌ sh : command not found	You tried to run Linux install commands on Windows.	Use the Windows Ollama installer from the download page.
⚠️ No backends available!	Ollama isn’t running or no model is pulled.	Run ollama pull phi3:mini, then restart the backend.
❌ uvicorn: command not found	FastAPI server not installed in the virtual environment.	Run pip install fastapi uvicorn httpx inside backend/venv.
⚠️ Static directory not found	You haven’t built the frontend yet.	Run npm run build inside frontend/.
🧱 EPERM: operation not permitted, stat 'postcss-nested'	Windows permission lock on node_modules.	Close VS Code, delete node_modules, run npm install again as Administrator.
😵 "Sorry, I encountered an error" in web UI	Frontend expected JSON but backend sent a stream.	Make sure your backend has the non-streaming fix enabled (see docs below).
🐢 Everything is slow	Running a big model on CPU.	Try phi3:mini instead of codellama:7b-instruct.

---

🧠 Tips for Low-Spec PCs

• Use phi3:mini or tinyllama:1.1b for best performance.
• Keep the FastAPI window open while using Hikari.
• Close browsers/tabs to free up RAM.
• To move models to another drive:

  mklink /D "C:\Users\<YourUser>\.ollama\models" "D:\LLMs\models"

---

🧩 Folder Structure

hikari/
 ├── backend/         # FastAPI + Ollama server
 │   ├── main.py
 │   ├── requirements.txt
 ├── frontend/        # Web UI built with Vite
 │   ├── dist/
 │   ├── package.json
 └── README.md

---

💡 Future Updates

• Real-time typing (streaming responses)
• More supported models
• Offline installer package
• UI themes (Rose Pine / Tokyo Night)

---

🪴 License

Released under the MIT License.
You’re free to use, modify, and share — just credit Houndslight / Hikari.

---

🌸 Created with love by Houndslight

“From us, to you.”