The M22 project ran a systematic, named feature-extraction program against multiple VPN apps — Express VPN, 7VPN, Master Nextgen, and Giti VPN — each ticket following the same workflow: extract signature in a lab environment, then run a false-positive ('误封') test in the Beijing environment before shipping to the field. Testing includes automated circumvention dial-testing (e.g. 17-hour looped Android tests).

Raw TSG session_record export logs dated Nov 2024, tagged device_group 'YGN-MYTEL' (Yangon, Mytel), directly corroborate the leak's M22=Myanmar/Mytel/Yangon site-codename mapping with live production data: individual SSL sessions are classified by app as 'Hotspot Shield VPN' (destination get.adobe.com, likely a fronting/CDN endpoint) and allowed under a named whitelist rule 'whitelist_102024', geolocated client-side as Myanmar.Yangon.Yangon.

TSG's app/protocol detection ("AppSketch" / context_based_detector plugin, part of the SAPP packet pipeline) is not purely static-signature: each detection rule can be an arbitrary Lua script (APP_SIG_LUA_SCRIPTS table) executed per-session in a per-worker-thread LuaJIT VM, with access to packet payload, session context counters, and helper functions (APP.data, APP.context.c2s_count, APP.log_debug, APP.append_extra_info). A validation CLI tool (luac-tool) checks script syntax, timeout, and return-value type before import.

Sustained, systematic feature-extraction campaign against ~20+ named commercial VPN apps (Secure VPN, VPN Turkey-series regional apps, Cloudbric, Thunder VPN, Quark VPN, TowerVPN, SymlexVPN, Tenta, tigerVPN, TLS Tunnel, Tomato VPN, VPNlat, VPNMonster, VPNPro, WhiteHatVPN, WolfVPN, and an ISAKMP/IKE-based and an OpenVPN-based app) for the Myanmar (M22) deployment, 2024. Methodology: reverse-engineer app API via Fiddler/mitm to enumerate server IP/FQDN lists where possible; fall back to Wireshark protocol-signature ID (e.g. recognizing raw ISAKMP port 500 or OpenVPN ports 1194/80) when APIs are unavailable; automated 'dial testing' (自动化拨测) continuously discovers new server IPs; ongoing re-verification explicitly tracks 'CT' (circumvention/breakthrough) to catch when a blocked app resumes working and needs a fingerprint update.

Geedge's M22 (Myanmar) team ran an active fingerprinting program specifically against "lanternVPN" in Aug 2024. They found the Android client's TLSv1 connections have fixed-position header fields (protocol version, payload length) usable as a static byte-pattern block signature, and that fake-SNI TLS 1.2/1.3 connections could only be blocked via destination-IP lists. They then built an automated pipeline (JA4 flow classification + domain/IP correlation + emulator-driven "new device" node discovery) that harvested 400-800+ new server IPs per week over the following month.

M22's extraction against "VPN Hero" (OpenVPN-based) recovered 13 FQDN + 13 IP indicators; FQDNs follow a "zampakuto...shop" template. Post-signature testing in the Myanmar demo environment found no successful connections on Android/iOS.

Documents the exact methodology used to derive a block signature for 'Giti VPN': TLS interception (fiddler) to decrypt the app's traffic, identifying its control/initialization domain and a second per-session domain pattern, then blocking by those two domain-name features rather than deep packet content.