Strix v0.1.3

Strix is a command-line penetration testing assistant powered by AI. It integrates with various AI models to help cybersecurity professionals and enthusiasts with technical tasks, while maintaining strict security controls to prevent unintended actions.

Features

• Multi-AI Support: Works with Google Gemini, OpenAI GPT-4, Anthropic Claude, Groq LLaMA 3, and Mistral
• Expanded Model Selection: Access to multiple models from each provider (e.g., gemini-2.5-flash, gpt-4o, claude-3-opus, etc.)
• Command-Line Model Selection: Direct model selection via --model command (new in v0.1.3)
• Tool Integration: Built-in functions for common pentesting tools (nmap, subfinder, gobuster, etc.)
• Interactive Mode: Real-time conversation interface with the AI assistant
• Enhanced UX: Improved user experience with prompt_toolkit for better input handling
• Confirmation Prompts: Asks for confirmation before executing dangerous commands
• Auto-Save Mode: Optional bypass for experienced users

What's New in v0.1.3

• New --model Command: Direct model selection from command line
  • Specify exact models: strix --model gpt-4, strix --model gemini-2.5-flash
  • Use provider names: strix --model openai, strix --model groq
  • Combine with other options: strix --model gpt-4 --prompt ctf --auto-save
• Enhanced Dependency Management: All AI provider packages now included in installation

What's New in v0.1.2

• Bug Fixes: Resolved several stability issues and bugs
• Performance Improvements: Faster responses and better overall performance
• UI Enhancements: Minor interface improvements

What's New in v0.1.1

• Expanded Model Support
  • Google Gemini: gemini-2.5-flash, gemini-2.0-flash, gemini-1.5-pro, gemini-1.5-pro-exp, gemini-1.0-pro
  • OpenAI: gpt-4, gpt-4-turbo, gpt-4o, gpt-3.5-turbo
  • Anthropic: claude-3-sonnet, claude-3-opus, claude-3-haiku, claude-2.1
  • Groq: llama3-70b-8192, llama-3.1-8b, llama-3.1-70b, mixtral-8x7b, gemma-7b
  • Mistral: mistral-small-latest, mistral-large, mistral-medium, mistral-nemo
• Enhanced User Experience: Improved input handling with prompt_toolkit
• Better Model Selection Menu

Installation

# Clone the repository
git clone https://github.com/strixproject/Strix.git
cd strix

# Option 1: Install in virtual environment (recommended)
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate
pip install -e . # Or pipx install -e .

# Option 2: Install globally/local (without virtual environment)
pip install -e . # Or pipx install -e .

# If the binary is in ~/.local/share/pipx/venvs/strix/bin/strix
sudo ln -sf ~/.local/share/pipx/venvs/strix/bin/strix /usr/local/bin/strix

# Set up API keys
# Create ~/Strix/.env with your API keys:
# GOOGLE_API_KEY=your_google_api_key_here
# OPENAI_API_KEY=your_openai_api_key_here
# ANTHROPIC_API_KEY=your_anthropic_api_key_here
# GROQ_API_KEY=your_groq_api_key_here
# MISTRAL_API_KEY=your_mistral_api_key_here

Usage

# Start interactive mode
strix

# Start with auto-save mode (bypass confirmation prompts)
strix --auto-save

# Start with a specific system prompt
strix --prompt ctf
strix --prompt vuln-research

# Start with a specific AI model (new in v0.1.3)
strix --model gpt-4
strix --model gemini-2.5-flash
strix --model claude-3-sonnet
strix --model llama3-70b-8192
strix --model mistral-large

# You can also specify provider names (uses default model for that provider)
strix --model openai      # Uses gpt-4 by default
strix --model gemini      # Uses gemini-2.5-flash by default
strix --model anthropic   # Uses claude-3-sonnet by default
strix --model groq        # Uses llama3-70b-8192 by default
strix --model mistral     # Uses mistral-small-latest by default

# Combine with other options
strix --model gpt-4 --prompt ctf --auto-save

# Show help
strix --help

System Prompts

Strix comes with three built-in system prompts that can be selected using the --prompt option:

Pentesting Assistant (Default)

strix --prompt pentest

Rules:
1. NEVER save scan results to a file automatically.
2. ONLY save to a file if the user explicitly asks (e.g., "save to file.txt" or "write this to output.txt").
3. When scanning (subdomains, ports, web), show output in the terminal only.
4. If the user provides a list and says "save to X", use 'write_file' with that content.
5. Be precise, technical, and do not hallucinate actions.
6. Match the user's language.
7. Use markdown: **bold**, *italic*, * lists.
8. You can create any script according to user requests, and can save the script via the save file function.

CTF Assistant

strix --prompt ctf

Rules:
1. Help users solve CTF challenges ethically and educationally.
2. Break down complex problems into understandable steps.
3. Explain methodologies and reasoning clearly.
4. Suggest multiple approaches when applicable.
5. Point out common pitfalls and how to avoid them.
6. Encourage learning and understanding over quick solutions.
7. Respect challenge categories (crypto, forensics, web, etc.).

Vulnerability Researcher

strix --prompt vuln-research

Rules:
1. Assist with vulnerability analysis and research methodologies.
2. NEVER exploit vulnerabilities in real systems without authorization.
3. Explain vulnerability concepts with practical examples.
4. Guide users through secure coding practices.
5. Help with CVE analysis and PoC development in controlled environments.
6. Provide guidance on responsible disclosure procedures.
7. Detail attack vectors and mitigation strategies.
8. Emphasize ethical considerations in all recommendations.

Supported AI Models

Google Gemini

• gemini-2.5-flash (default)
• gemini-2.0-flash
• gemini-1.5-pro
• gemini-1.5-pro-exp
• gemini-1.0-pro

OpenAI

• gpt-4
• gpt-4-turbo
• gpt-4o
• gpt-3.5-turbo

Anthropic Claude

• claude-3-sonnet
• claude-3-opus
• claude-3-haiku
• claude-2.1

Groq

• llama3-70b-8192
• llama-3.1-8b
• llama-3.1-70b
• mixtral-8x7b
• gemma-7b

Mistral

• mistral-small-latest
• mistral-large
• mistral-medium
• mistral-nemo

Required Tools

For full functionality, install these security tools:

• subfinder
• nmap
• gobuster Or you can use your local tools that are already installed.

License

MIT License - see LICENSE file for details.

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.