jfabian/m3sh
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
Lightweight network orchestrator for building non-persistent, high-entropy honeypots. Using a seed-based MIND engine, it emulates realistic lateral movement scenarios (Windows/Linux/Cisco) to trap attackers and train autonomous security agents in a zero-footprint environment.
1 commit 1 branch 0 tags 35 KiB
  • Rust 90.4%
  • Makefile 9.6%
Find a file
Exact
jfabian 172851c77e feat(mind): implement Point Zero procedural infrastructure engine 
- Migrated from static JSON-based generation to a stateless Rust MIND engine.
- Implemented deterministic network projection using PCG64 seed-based RNG.
- Developed a collision-free IP allocation algorithm (GLC) for /24 subnets.
- Integrated YAML-blueprint loading for scalable asset emulation.
- Added initial Asset Library: Cisco IOS, Windows Server, pfSense, and Linux.
- Established a high-fidelity environment for Security Specialists and AI Agents.
2026-04-10 11:38:44 -03:00
blueprints feat(mind): implement Point Zero procedural infrastructure engine 2026-04-10 11:38:44 -03:00
src feat(mind): implement Point Zero procedural infrastructure engine 2026-04-10 11:38:44 -03:00
.gitignore feat(mind): implement Point Zero procedural infrastructure engine 2026-04-10 11:38:44 -03:00
Cargo.lock feat(mind): implement Point Zero procedural infrastructure engine 2026-04-10 11:38:44 -03:00
Cargo.toml feat(mind): implement Point Zero procedural infrastructure engine 2026-04-10 11:38:44 -03:00
Makefile feat(mind): implement Point Zero procedural infrastructure engine 2026-04-10 11:38:44 -03:00
README.md feat(mind): implement Point Zero procedural infrastructure engine 2026-04-10 11:38:44 -03:00

README.md

🧠 M3[SH]: The Virtual Network Emulator

M3[SH] is a lightweight, procedural network emulation framework built in Rust. It is designed for Detection Engineering, Red Teaming, and Deception strategy validation without the infrastructure overhead of traditional virtualization.

🎯 Key Objectives

  • Zero-Footprint Emulation: Generate complex, high-fidelity network topologies (Windows, Linux, Cisco) on-the-fly without the resource consumption of VMs or containers.
  • Procedural Reality: Using a Roguelike approach to infrastructure, every session is generated from a single seed. Analysts and Autonomous Agents must perform real reconnaissance as the network map is unique to each "Dream".
  • Deterministic Sovereignty: Built with a "local-first" philosophy, the MIND engine ensures that the entire network state is calculated mathematically, preventing data leakage.
  • Agentic Playground: Provides a consistent, scriptable, and low-latency environment for testing and training AI Security Agents in complex lateral movement scenarios.

🛠 Tech Stack

  • Language: Rust (Edition 2024).
  • Deterministic RNG: rand_pcg for replayable infrastructure projections.
  • Configuration: YAML-based blueprints for OS and service emulation.
  • Core Logic: The MIND engine, responsible for stateless IP mapping (GLC) and node projection.

🚀 Getting Started (Point Zero)

Currently, M3[SH] can project its internal "Mind" to verify consistency.

  1. Clone and build: 
    cargo build
  2. Run the projection: 
    cargo run

🗺 Project Roadmap

  • [DONE] Point Zero: Procedural engine, blueprint system, and deterministic IP allocation.
  • [NEXT] The Gateway: Implementation of an asynchronous SSH server to allow real interaction with projected nodes.
  • [PLANNED] The Nightmare: Introduction of network artifacts, latency simulation, and dynamic file systems.