QGo

# QGo — The Most Advanced AI Coding Assistant **Created by [Rahul Chaube](https://github.com/Rahulchaube1)** **More powerful than Cursor. Smarter than Aider. Multi-Agent AI that lives in your terminal.** [![CI](https://github.com/Rahulchaube1/QGo/actions/workflows/ci.yml/badge.svg)](https://github.com/Rahulchaube1/QGo/actions/workflows/ci.yml) [![Python 3.10+](https://img.shields.io/badge/python-3.10+-blue.svg)](https://www.python.org/) [![License: Apache-2.0](https://img.shields.io/badge/License-Apache--2.0-green.svg)](LICENSE) [![Models: 50+](https://img.shields.io/badge/models-50+-purple.svg)](#-supported-models) [![Agents: 8](https://img.shields.io/badge/agents-8-orange.svg)](#-multi-agent-system) [![Copyright: Rahul Chaube](https://img.shields.io/badge/copyright-Rahul%20Chaube-blueviolet.svg)](#-copyright) > *"The AI coding assistant that thinks like a team of senior engineers."* > — Rahul Chaube

✨ Features

Feature	Description
🤖 Multi-LLM	OpenAI, Anthropic Claude, DeepSeek, Gemini, Groq, Mistral, Ollama (local), and any litellm-supported model
🗺️ Repo Map	Automatically maps your entire codebase so the AI understands structure and context
✏️ Smart Editing	4 editing formats: SEARCH/REPLACE blocks, whole-file, unified diffs, architect mode
🏗️ Architect Mode	Two-pass: strong model plans → fast model implements
👥 Multi-Agent System	8 specialist agents (Planner, Coder, Reviewer, Tester, Debugger, DocWriter, Security, Refactor) work collaboratively
🔀 Git Integration	Auto-commits, diffs, undo, branch management
🎨 Beautiful UI	Rich terminal output with syntax highlighting, Markdown rendering, live streaming
💬 Interactive REPL	Full-featured REPL with history, tab completion, multi-line input
🌐 Web Context	Fetch any URL and add it as context (`/web https://...`)
📎 Clipboard	Paste clipboard content into the conversation (`/paste`)
🔍 Auto Lint	Automatically run your linter after each edit; ask AI to fix errors
🧪 Auto Test	Run your test suite and fix failures automatically
🔒 Security Audit	Dedicated security agent audits your code for vulnerabilities
⚙️ Configurable	Project-level `.qgo.conf` + user-level `~/.qgo.conf` + env vars
🔌 Local Models	Full Ollama support — run llama3, codellama, mistral, qwen locally
📦 100+ Languages	Python, JS/TS, Go, Rust, Java, C/C++, Ruby, PHP, Swift, Kotlin, and more

👥 Multi-Agent System

QGo’s multi-agent system is what sets it apart. Instead of a single AI trying to do everything, QGo orchestrates a team of 8 specialist agents that collaborate like a real engineering team:

User Request
     │
     ▼
┌─────────────┐     Plans sub-tasks
│ 📋 PLANNER  │──────────────────────────────────────────────┐
└─────────────┘                                              │
                                                             ▼
┌─────────┐  ┌──────────┐  ┌────────┐  ┌──────────┐  ┌──────────┐
│⚙️ CODER │  │🔍 REVIEW │  │🧪 TEST │  │🐛 DEBUG  │  │🔒 SECURI-│
│         │  │          │  │        │  │          │  │   TY     │
└────┬────┘  └────┬─────┘  └───┬────┘  └────┬─────┘  └────┬─────┘
     │             │            │             │              │
     └─────────────┴────────────┴─────────────┴──────────────┘
                                     │
                              ┌──────▼──────┐
                              │ 📝 DOC +    │
                              │ ♻️ REFACTOR  │
                              └──────┬──────┘
                                     │
                              ┌──────▼──────┐
                              │  Final      │
                              │  Report     │
                              └─────────────┘

Agent	Icon	Speciality
Planner	📋	Decomposes tasks into ordered sub-tasks, assigns to specialists
Coder	⚙️	Writes clean, production-quality code with type hints
Reviewer	🔍	Reviews for correctness, performance, and best practices
Tester	🧪	Writes comprehensive pytest suites with edge cases
Debugger	🐛	Root-cause analysis and minimal bug fixes
DocWriter	📝	Google-style docstrings, README sections, inline comments
Security	🔒	Audits for injections, path traversal, exposed secrets, CVEs
Refactor	♻️	Improves structure (DRY, SRP, naming) without breaking behaviour

Multi-Agent Usage

# Run full multi-agent pipeline (auto-plans and executes)
qgo agent "add authentication to the Flask API"

# Run a specific agent directly
qgo agent --agent coder "implement pagination for list_users()"
qgo agent --agent security "audit auth.py for vulnerabilities"
qgo agent --agent tester "write tests for the payment module"

# Interactive: use /agent command in REPL
[QGo] > /agent add rate limiting to the API

🚀 Installation

pip install qgo

Or install from source for the latest features:

git clone https://github.com/Rahulchaube1/QGo
cd QGo
pip install -e .

⚡ Quick Start

# Interactive session (most common)
qgo

# One-shot command
qgo "add docstrings to all functions in utils.py"

# Add specific files to context
qgo --file main.py --file utils.py "refactor the error handling"

# Use a specific model
qgo --model claude-3-7-sonnet-20250219
qgo --model deepseek/deepseek-chat
qgo --model ollama/llama3.2 --api-base http://localhost:11434

🔑 Setting Your API Key

# OpenAI
export OPENAI_API_KEY=sk-...

# Anthropic
export ANTHROPIC_API_KEY=sk-ant-...

# DeepSeek
export DEEPSEEK_API_KEY=sk-...

# Google Gemini
export GEMINI_API_KEY=...

# Or pass directly
qgo --api-key sk-... --model gpt-4o

💻 Usage

Interactive Mode

[QGo] > /add src/main.py src/utils.py
✓ Added: src/main.py
✓ Added: src/utils.py

[QGo] > add type hints to all functions

QGo is thinking...

# ... AI edits the files with SEARCH/REPLACE blocks ...
📦 Committed: a1b2c3d4  add type hints to all functions

[QGo] > /diff
# shows the diff...

[QGo] > /undo
✓ Undid last commit (changes kept in working tree)

One-Shot Mode

# Fix a bug
qgo --file main.py "fix the null pointer exception in parse_config()"

# Add a feature
qgo --file api.py --file models.py "add pagination to the list_users endpoint"

# Write tests
qgo --file calculator.py "write comprehensive pytest tests for all functions"

# Explain code
qgo --file complex_algo.py "explain what this algorithm does"

Architect Mode (for large tasks)

# Uses a strong model to plan, then a fast model to implement
qgo --edit-format architect "implement a REST API with authentication"

📝 Edit Formats

Format	Flag	Best For
editblock	`--edit-format editblock`	Default. Most reliable. SEARCH/REPLACE blocks
whole	`--edit-format whole`	Small files. Complete file replacement
udiff	`--edit-format udiff`	When you need standard unified diffs
architect	`--edit-format architect`	Complex multi-file changes. Two-pass approach

🤖 Supported Models

Provider	Models
OpenAI	gpt-4o, gpt-4o-mini, gpt-4-turbo, o1, o1-mini, o3-mini
Anthropic	claude-3-7-sonnet, claude-3-5-sonnet, claude-3-opus, claude-3-haiku
DeepSeek	deepseek-chat, deepseek-coder, deepseek-r1
Google	gemini-1.5-pro, gemini-1.5-flash, gemini-2.0-flash
Groq	llama3-8b, llama3-70b, mixtral-8x7b (ultra-fast)
Mistral	mistral-large, codestral
Cohere	command-r, command-r-plus
Ollama	llama3.2, codellama, mistral, qwen2.5-coder, deepseek-coder-v2 (local)

# List all models
qgo models

# List by provider
qgo models --provider anthropic

🎮 REPL Commands

Command	Description
`/add <files>`	Add files to context (supports globs: `/add src/*.py`)
`/drop <files>`	Remove files from context
`/files`	List files currently in context
`/diff`	Show uncommitted git diff
`/commit [msg]`	Manually commit changes
`/undo`	Undo last git commit (keeps changes)
`/clear`	Clear conversation history
`/model <name>`	Switch to a different model
`/models`	List all available models
`/tokens`	Show token usage estimate
`/map`	Show repository map
`/run <cmd>`	Run a shell command, add output to context
`/web <url>`	Fetch a URL and add content to context
`/git <cmd>`	Run any git command
`/paste`	Paste clipboard content into conversation
`/ls [path]`	List directory contents
`/config`	Show current configuration
`/help`	Show help
`/exit`	Exit QGo

⚙️ Configuration

QGo reads configuration from (highest priority first):

CLI flags
Environment variables (QGO_*)
Project-level .qgo.conf
User-level ~/.qgo.conf

`.qgo.conf` Example

model: claude-3-7-sonnet-20250219
edit_format: editblock
auto_commits: true
show_diffs: true
map_tokens: 2048
auto_lint: true
lint_cmd: ruff check
auto_test: false
test_cmd: pytest
stream: true

Environment Variables

Variable	Description
`QGO_MODEL`	Default model
`QGO_API_KEY`	API key
`QGO_API_BASE`	Custom API base URL
`QGO_EDIT_FORMAT`	Edit format
`QGO_AUTO_COMMITS`	Enable/disable auto-commits
`QGO_MAP_TOKENS`	Max tokens for repo map
`QGO_AUTO_LINT`	Enable auto-linting
`QGO_LINT_CMD`	Linter command
`QGO_AUTO_TEST`	Enable auto-testing
`QGO_TEST_CMD`	Test command

🏗️ Architecture

qgo/
├── agents/                 # Multi-agent orchestration system
│   ├── orchestrator.py     # Coordinates all agents (AgentOrchestrator)
│   ├── base_agent.py       # Abstract BaseAgent, AgentMessage, AgentResult
│   └── specialist_agents.py # Planner, Coder, Reviewer, Tester, Debugger, DocWriter, Security, Refactor
├── llm/                    # Universal LLM backends
│   ├── litellm_provider.py # 100+ models via litellm
│   ├── model_info.py       # Model metadata & costs
│   └── streaming.py        # Streaming response handling
├── coders/                 # Code editing engines
│   ├── base_coder.py       # Core chat+edit loop
│   ├── editblock_coder.py  # SEARCH/REPLACE (default)
│   ├── whole_coder.py      # Full file replacement
│   ├── udiff_coder.py      # Unified diff
│   └── architect_coder.py  # Two-pass plan+implement
├── repo/                   # Repository understanding
│   ├── repo_map.py         # Codebase symbol mapping
│   ├── git_repo.py         # Git operations
│   └── file_watcher.py     # Live file watching
├── ui/                     # Terminal user interface
│   ├── terminal.py         # Rich-based I/O
│   ├── repl.py             # Interactive REPL
│   └── commands.py         # Slash-command handler
├── utils/                  # Utilities
│   ├── file_utils.py       # File operations + diffs
│   ├── token_counter.py    # tiktoken-based counting
│   └── web_scraper.py      # URL fetching
├── config.py               # Configuration management
├── models.py               # Data types
└── main.py                 # CLI entry point

🔧 Development

# Clone and install dev dependencies
git clone https://github.com/Rahulchaube1/QGo
cd QGo
pip install -e ".[dev]"

# Run tests
pytest tests/ -v

# Lint
ruff check qgo/

# Format
black qgo/ tests/

🤝 Contributing

Contributions are welcome! Please:

Fork the repository
Create a feature branch
Add tests for new features
Run ruff check and pytest before submitting
Open a pull request

📄 License

Apache 2.0 — see LICENSE for details.

© Copyright

QGo is created and maintained by Rahul Chaube.

Copyright (c) 2024 Rahul Chaube. All Rights Reserved.

Licensed under the Apache License, Version 2.0.
You may obtain a copy of the License at:
    http://www.apache.org/licenses/LICENSE-2.0

Author:  Rahul Chaube
GitHub:  https://github.com/Rahulchaube1
Project: https://github.com/Rahulchaube1/QGo

All source files, documentation, assets, and configuration in this repository are the intellectual property of Rahul Chaube. The QGo name, logo, and branding are owned by Rahul Chaube.

QGo