C Code Assistant

A fully local AI-powered C programming assistant.
No cloud. No API keys. Runs entirely on your machine via Ollama.

---

What it does

C Code Assistant is a browser-based IDE companion for C programming, powered by locally running LLMs. It can complete, generate, explain, debug, and optimize C code — all without sending anything to the internet.

All Ollama traffic is routed through the Flask backend (localhost proxy), so Ollama never needs to be exposed to any external origin.

---

Features

Feature	Description
⚡ Complete	Paste partial C code, get it finished
✨ Generate	Describe in plain English, get working C
📖 Explain	Understand any C code in plain English
🔧 Fix / Debug	Finds and fixes every bug with explanations
🚀 Optimize	Improves performance with Big-O analysis
▶ Compile & Run	Compiles with GCC and runs in the browser
🔍 Lint	Live linting with cppcheck or GCC fallback
🔄 Quality Retry	Output is auto-checked and rewritten if wrong
🤖 Model Switcher	Switch between any installed Ollama model
🕘 History	Session history saved in localStorage
⬇ Export .c	Download code as a .c file

---

Requirements

Required

• Python 3.8+
• Ollama with at least one model pulled

Optional (for Compile & Run)

• GCC
  • Windows: MinGW-w64 — download, extract, add bin/ to PATH
  • Linux: sudo apt install gcc
  • macOS: xcode-select --install
• cppcheck (richer linting)
  • Windows: cppcheck.sourceforge.io
  • Linux: sudo apt install cppcheck
  • macOS: brew install cppcheck

---

Recommended Models

Model	Size	Pull Command
qwen2.5-coder:1.5b	~1 GB	ollama pull qwen2.5-coder:1.5b
deepseek-coder:1.3b	~776 MB	ollama pull deepseek-coder:1.3b
qwen2.5-coder:7b	~4.7 GB	ollama pull qwen2.5-coder:7b
deepseek-coder:6.7b	~3.8 GB	ollama pull deepseek-coder:6.7b

---

Installation

1. Clone the repo

git clone https://github.com/CODExGAMERZ/C-Code-Assistant.git
cd c-code-assistant

2. Install Python dependencies

python -m pip install flask flask-cors requests

3. Pull a model

ollama pull qwen2.5-coder:1.5b

4. Start the app

Windows — double-click start.bat or run:

python server.py

Linux / macOS:

bash start.sh

5. Open in browser

http://localhost:5050

⚠️ Always open http://localhost:5050 — never open index.html directly as a file.

---

Architecture

Browser (localhost:5050)
    │
    ├── GET  /              → serves index.html
    ├── GET  /js/*.js       → serves JS files
    ├── GET  /css/*.css     → serves CSS files
    │
    ├── POST /api/compile   → GCC compile & run
    ├── POST /api/lint      → cppcheck / GCC lint
    │
    └── /api/ollama/*  ──proxy──▶  Ollama (localhost:11434)
            ├── GET  /api/ollama/tags
            ├── POST /api/ollama/chat      (streaming)
            └── POST /api/ollama/generate

The browser never contacts Ollama directly. All LLM traffic goes through the Flask proxy, so Ollama runs with default settings — no OLLAMA_ORIGINS flag required.

---

Quality-Retry System

Every response goes through an automatic quality gate:

User clicks Run
      │
      ▼
LLM generates output (streamed live)
      │
      ▼
Quality checker reviews:
  - Valid C code in fenced block?
  - No unresolved TODO / placeholder?
  - No type errors (int passed as char*)?
  - All functions complete?
      │
   pass ──▶ Show output  ● green
      │
   fail ──▶ Retry with failure reason injected
            (up to 2 retries, configurable)
              │
              ▼
           Show best result  ● amber

---

Project Structure

c-code-assistant/
├── index.html         ← App UI
├── server.py          ← Flask backend + Ollama proxy
├── start.bat          ← Windows launcher
├── start.sh           ← Linux/macOS launcher
├── LICENSE
├── README.md
├── .gitignore
├── css/
│   └── style.css      ← All styles
└── js/
    ├── config.js      ← Settings & model switcher
    ├── api.js         ← Ollama proxy calls + quality retry
    ├── app.js         ← Main controller
    ├── compiler.js    ← GCC compile & run
    ├── linter.js      ← Live linting
    ├── editor.js      ← Code editor (line nums, tab, indent)
    ├── output.js      ← C syntax highlighter + renderer
    └── history.js     ← Session history

---

Keyboard Shortcuts

Shortcut	Action
Ctrl+Enter	Run
Ctrl+L	Clear editor
Ctrl+S	Export as .c file
Tab	Indent 4 spaces
Shift+Tab	De-indent

---

Troubleshooting

Ollama shows offline

• Make sure Ollama is running: ollama serve
• The app auto-retries every 12 seconds

Model not found in dropdown

• Pull it: ollama pull <model-name>
• Or select a model you already have

Compile & Run not working

• Install GCC and add it to PATH
• Test with: gcc --version in a terminal

Output has bugs / wrong code

• Try a larger model (qwen2.5-coder:7b)
• Lower temperature to 0.1 in settings
• Increase max retries to 3 in settings

---

Acknowledgements

• Ollama — local LLM runtime
• Qwen2.5-Coder — by Alibaba Cloud
• DeepSeek-Coder — by DeepSeek AI
• Flask — Python web framework
• cppcheck — C/C++ static analysis
• IBM Plex Mono — editor font

---

License

MIT — see LICENSE