This project provides a suite of Claude Code SubAgents designed to automate and streamline the application of code evasion techniques. It targets security researchers and developers seeking to enhance the stealth capabilities of loaders, C2 frameworks, and penetration testing tools by leveraging Claude Code's AI-assisted development environment. The primary benefit is the automation of complex, manual tasks involved in researching, generating, and integrating sophisticated evasion strategies.

How It Works

This project leverages Claude Code's plugin architecture, organizing functionality into distinct SubAgents and Skills. User commands trigger agents that interact with a structured knowledge base (evasion_techniques.json, loader_techniques.json) managed by lib/knowledge_manager.py for research, generation, and integration tasks. The core agents include research-agent for GitHub analysis, loadergen-agent for generating shellcode loaders, evasion-agent for integrating evasion into loaders, c2-evasion-agent for C2 framework modification, and tools-evasion-agent for penetration testing tool evasion. This approach automates complex security research and development workflows for code evasion.

Quick Start & Requirements

• Primary Install/Run: Install Claude Code globally (npm install -g @anthropic-ai/claude-code). Navigate to the project directory (cd evasion-agent-teams) and run claude.
• Plugin Loading: Recommended via project directory execution; alternatively, configure ~/.claude/CLAUDE.md globally.
• Prerequisites: Python 3.8+, MinGW-w64 (latest), GitHub CLI (gh) 2.0+, Node.js. Setup involves installing these tools and configuring PATH variables.
• Links: Claude Code documentation: https://github.com/anthropics/claude-code. Python downloads: https://www.python.org/downloads/.

Highlighted Details

• Dedicated agents for security research, loader generation, evasion integration, C2 framework evasion, and penetration testing tool evasion.
• Centralized knowledge base management via Python scripts for evasion techniques, loader components, and scenario tracking.
• Extensive command-line interface for research, generation, integration, and analysis tasks.
• Specific safety rules are defined for each agent, restricting potentially harmful operations.

Maintenance & Community

No specific details regarding maintainers, community channels (e.g., Discord/Slack), or project roadmap are provided in the README.

Licensing & Compatibility

• License Type: MIT License.
• Compatibility: The MIT license is permissive and generally compatible with commercial use and closed-source linking.

Limitations & Caveats

Agent security policies restrict the execution of external code and mandate careful modification of network tool fields to preserve length. Setup complexity arises from multiple development tool dependencies. Functionality is entirely dependent on the Claude Code framework. Testing relies on specific shellcode formats (x64 Windows binary) and can fail if these are not met.