← all repositories
Ed1s0nZ/CyberStrikeAI

Go-based pentest platform wraps 100+ tools behind AI agents

CyberStrikeAI turns scattered security scanners into an orchestrated, chat-driven testing environment with a built-in C2 framework.

4.2k stars Go Domain AppsAgentsOther AI
CyberStrikeAI
Velocity · 7d
+20
★ / day
Trend
steady
star history

What it does CyberStrikeAI is a Go application that bundles 100+ security tools—nmap, sqlmap, nuclei, metasploit, and dozens more—into a single web console. You configure an OpenAI-compatible API key, pick a role (penetration tester, CTF player, web app scanner), and drive the whole stack through conversational commands or a chatbot. It handles vulnerability tracking, task queuing, knowledge-base retrieval, and even webshell and C2 session management in one place.

The interesting bit The project doesn’t just shell out to tools; it builds an agent layer on top of CloudWeGo’s Eino framework with three multi-agent orchestration modes—deep, plan-execute, and supervisor—plus progressive skill loading so agents only pull in relevant capabilities mid-session. There’s also a lightweight, encrypted C2 framework with listeners, implants, and MCP-exposed task queues, explicitly gated behind human-in-the-loop approval and “authorized testing only” warnings.

Key highlights

  • Native MCP server with HTTP, stdio, and SSE transports; federates with external MCP servers
  • 20+ sample skill packs (SQLi, XSS, API security) using Eino’s ADK skill tool format
  • Built-in RAG knowledge base with vector retrieval and optional Eino Compose indexing
  • Project-level “shared facts” blackboard that auto-injects context across sessions and agents
  • Burp Suite plugin, DingTalk/Lark chatbot integrations, and role-based tool restrictions
  • SQLite persistence, audit logs, and password-protected web UI with self-signed TLS by default

Caveats

  • Requires both Go 1.21+ and Python 3.10+; the run.sh bootstrap creates a Python venv and builds from source
  • Defaults to self-signed HTTPS on localhost; production TLS needs manual certificate configuration
  • README notes that AI will “automatically fall back to alternatives when a tool is missing,” but doesn’t clarify how reliably that works in practice

Verdict Worth a look if you’re running a security team that wants to consolidate tool sprawl behind a single, auditable interface with agentic automation. Skip it if you need a mature, vendor-supported platform or aren’t comfortable self-hosting a tool that includes its own C2 infrastructure.

heatdrop uses Google Analytics to see which pages get read — nothing else. Your call. How we handle data.