National Cyber Warfare Foundation (NCWF)login

MESHforensics for censorship-resistant remote mobile forensics over an encrypted mesh


0 user ratings
2026-11-11 19:31:22
milo
Red Team (CNA)
"MESHforensics

MESHforensics builds an encrypted, peer-to-peer mesh overlay so authorized analysts can run ADB wireless debugging and libimobiledevice collection on devices stuck behind hostile networks.








ToolBARGHEST-ngo/MESHforensics — encrypted, censorship-resistant mesh overlay enabling remote mobile forensics and network monitoring
CategoryMobile forensics / mesh VPN / DFIR transport
Primary UseRunning ADB Wireless Debugging, libimobiledevice, MVT, AndroidQF, and Suricata collection against consented devices in the field over an ephemeral WireGuard-based mesh
Safe UseDesigned for consensual remote forensics of spyware targeting Human Rights Defenders; intended for authorized DFIR practitioners, NGO investigators, and lab environments with device-owner consent
Telemetry NoteEphemeral task-scoped meshes are designed to leave minimal persistent infrastructure to fingerprint; UDP hole-punch attempts and HTTPS DERP fallback traffic are network-observable, and analyst-operated DERP servers keep relay visibility under the defender's control

MESHforensics, published by the BARGHEST NGO at BARGHEST-ngo/MESHforensics, attacks a stubborn operational problem in mobile DFIR: the device you need to examine is sitting behind NAT, a carrier-grade firewall, or a network that deliberately degrades UDP, while the analyst is somewhere else entirely. Rather than asking anyone to open ports or stand up permanent VPN infrastructure, the tool constructs an encrypted peer-to-peer overlay in which every node receives a CGNAT-range address on a virtual TUN interface. From the analyst's perspective, the remote phone behaves as though it were on the same private subnet as the workstation, which means the entire familiar forensic toolchain — ADB over WiFi, libimobiledevice, AndroidQF, MVT — works as if the device were plugged in locally. The repository reports 211 stars, is written in Kotlin on the client side, and is licensed AGPL-3.0-or-later.


The architectural detail that will interest most readers is buried matter-of-factly in the README: MESHforensics is a heavily modified fork of Tailscale with the Tailscale infrastructure requirement removed. Peers negotiate direct end-to-end encrypted WireGuard tunnels via UDP hole punching, and when UDP is blocked — which the authors correctly note is the first thing censored carrier networks kill — traffic fails over to end-to-end encrypted HTTPS DERP relays. Because DERP rides inside ordinary-looking HTTPS, it survives DPI that would strand a raw WireGuard deployment, and the relay still cannot read packet contents. The control plane, which the project also lets you self-host, handles only peer discovery and key exchange and carries no forensic traffic at all.


This separation of coordination from transport is the core design thesis. The README contrasts MESHforensics against hub-and-spoke VPN designs, which create persistent infrastructure to fingerprint, centralize traffic analysis, and add single points of failure. Here the mesh is ephemeral and analyst-controlled: you bring devices online, run your collection, and tear the network down so nothing is left listening. For high-risk environments — the project is funded by the Open Technology Fund explicitly to support consensual remote forensics of spyware targeting Human Rights Defenders — that disposability is not a convenience feature, it is a safety property.


On top of the transport, the project layers capability that maps directly onto established DFIR workflows. Automatic WireGuard and AmneziaWG key management is built in, with optional transport obfuscation against DPI. Remote network monitoring is supported over the same mesh, including PCAP capture and Suricata intrusion detection. Artifact transfer is described as secure, and there is an optional kill-switch containment mode for isolating a device during collection. The README also highlights integration points with sibling and third-party tools: WARD from the same organization, plus MVT and AndroidQF from the MVT project, all reachable as if local once the mesh is up.


Getting started is a two-path affair, and the choice is about how the control plane is reachable from the internet. Option A uses the MESHforensics provisioner, a hosted reverse-proxy tunnel that auto-assigns an ephemeral *.tunnels.meshforensics.app subdomain and issues API credentials, so the task apikey step is unnecessary. Option B is fully self-hosted: you bring your own ingress — a reverse proxy on an internet-facing box, a Cloudflare Tunnel, or ngrok — then mint the key locally. A minimal setup looks like git clone https://github.com/BARGHEST-ngo/mesh.git followed by task build and task controlPlane, choosing the Ephemeral or Persistent control-plane mode when prompted. The web UI ships with a self-signed certificate by default and is served on https://localhost or your exposed domain on port 8443.


One configuration warning in the README deserves emphasis because it inverts the usual secure default: the out-of-the-box ACL permits all nodes in a network to talk to each other. For any engagement involving more than one seized or consented device, that means forensic artifacts from one target could be reachable from another node. The authors direct production deployments to the ACL tab to apply restrictive policies, and any professional deploying this in an authorized lab should treat that step as mandatory before enrolling the first device, both for evidentiary hygiene and for containment of potentially compromised handsets.


The repository is organized into three components: android-client (the Android endpoint APK), control-plane (the coordination server), and analyst (the CLI client). Developer notes indicate a disciplined process — branch-based development merged through pull requests, versioned tag releases, and a GitHub Actions pipeline that mirrors tagged releases to a mesh-analyst-client distribution repository. External Go projects are told to depend on explicit version tags rather than main, which is sensible guidance for anyone embedding the mesh into a larger forensic pipeline. The license situation is worth reading too: the project is AGPL-3.0-or-later overall, with derivative Tailscale portions under BSD-3-Clause, AmneziaWG/WireGuard code under MIT, and original notices preserved in .licenses/.


Maturity is candidly labeled: the project is in public beta, under active development, with breaking changes expected. Notably, the README states that a full penetration test has been completed and major vulnerabilities patched — an unusually concrete claim for a tool at this stage, and a signal that BARGHEST has treated the security of the forensic transport itself as part of the threat model. Bug and security reports are routed through GitHub Issues, and full documentation lives at docs.meshforensics.org.


Where does this fit relative to just running plain WireGuard? The README's own argument is convincing: WireGuard speaks only UDP, and UDP is precisely what restrictive networks degrade first, so a conventional tunnel fails exactly when the analyst needs it most. MESHforensics keeps WireGuard as the fast path, adds hole-punching for NAT traversal, and treats HTTPS DERP relaying as the survival mechanism, with the option to run your own DERP server embedded in the control plane for a fully self-hosted deployment. Operators can also fall back to Tailscale's public DERP servers for connectivity, a trade-off worth weighing when relay metadata sensitivity matters.


For defenders on the blue-team side, the telemetry profile cuts both ways and is worth understanding. The ephemeral mesh design intentionally leaves little standing infrastructure to fingerprint, and direct peer connections carry no relay-visible content. At the same time, an authorized SOC observing its own network would see UDP hole-punch attempts, WireGuard/AmneziaWG handshakes or their obfuscated variants, and — where UDP is blocked — persistent HTTPS sessions to DERP endpoints that are indistinguishable from ordinary web traffic. Running analyst-controlled DERP servers keeps that relay visibility inside the investigation rather than leaking it to third parties.


The intended use case deserves the final word because it frames everything else. This is a tool built for consensual, authorization-first forensics: an Human Rights Defender who suspects stalkerware or commercial spyware on their phone, working with an NGO analyst who needs to run MVT-style checks without physical access, over networks that may be actively hostile to both the analyst and the subject. Everything in the design — censorship resistance, ephemeral topology, end-to-end encryption, kill-switch containment, self-hostability — serves that mission. Deployed inside that authorized context, MESHforensics is one of the more thoughtful recent entries in mobile DFIR transport; deployed outside it, the same properties would make it a network-access question worth answering with the ACL hygiene and telemetry angles above.



Official project repository for BARGHEST-ngo/MESHforensics.

Download Tool

Educational analysis for authorized security professionals. Use only in controlled, authorized environments.






Source: OffensiveSec
Source Link: https://www.offsecblog.com/2026/11/meshforensics-for-censorship-resistant.html


Comments
new comment
Nobody has commented yet. Will you be the first?
 
Forum
Red Team (CNA)



Copyright 2012 through 2026 - National Cyber Warfare Foundation - All rights reserved worldwide.