AI-native penetration-testing copilot built in Go, CyberStrikeAI automates security engagements by integrating hundreds of security tools with an AI decision engine and the MCP protocol. It allows users to run full penetration tests through natural language conversations, simplifying complex security testing workflows for engineers and researchers.

How It Works

CyberStrikeAI employs an AI decision engine that interfaces with OpenAI-compatible models (e.g., GPT, Claude, DeepSeek) to interpret findings and orchestrate tool execution. It utilizes a native MCP (Meta-Control Protocol) implementation supporting HTTP and stdio transports, with capabilities for external MCP server federation. The platform's extensibility is powered by a YAML-based system for defining and customizing over 100 prebuilt security tool recipes, featuring automatic fallback mechanisms for missing tools.

Quick Start & Requirements

• Installation: Clone the repository (git clone https://github.com/Ed1s0nZ/CyberStrikeAI.git), navigate into the directory, and run go mod download.
• Configuration: Set up OpenAI-compatible API access (key, base URL, model) in config.yaml or via the in-app Settings panel. Configure authentication password (or leave empty for auto-generation).
• Prerequisites: Go programming language. Optional: macOS (brew) or Ubuntu/Debian (apt-get) for installing common security tools (e.g., nmap, sqlmap, nuclei). The system will attempt to use alternative tools if specific ones are not installed.
• Launch: Execute ./run.sh, go run cmd/server/main.go, or build and run the binary (go build -o cyberstrike-ai cmd/server/main.go).
• Access: The web console is accessible at http://localhost:8080.
• Repository: https://github.com/Ed1s0nZ/CyberStrikeAI.git

Highlighted Details

• AI decision engine supporting various OpenAI-compatible models for intelligent orchestration.
• Native MCP implementation with HTTP/stdio transports and external MCP federation capabilities.
• Extensible tool system with 100+ prebuilt YAML recipes, allowing custom tool integration.
• Advanced handling of large tool outputs via pagination, compression, and searchable archives.
• Interactive attack-chain visualization with risk scoring and step-by-step replay functionality.
• Secure web UI with password protection, audit logs, and SQLite persistence.

Maintenance & Community

Community contributions for tooling are welcomed via issues or pull requests.

Licensing & Compatibility

No specific license information is provided in the README.

Limitations & Caveats

The platform relies on the availability of specific tools, with AI falling back to alternatives when required tools are missing, which may impact execution completeness. Custom tool integration requires adherence to the YAML recipe schema.