Flowerbed
- Type
- Backdoor
- Capabilities
- ExfiltrationPost exploitation
- Platforms
- Cloud
Flowerbed is a Python-based, Docker-deployed server-side collection framework used by the Russian state-sponsored espionage actor LAUNDRY BEAR, also tracked as Void Blizzard, CL-STA-1114, and TA488. It supports post-compromise data collection and receipt of exfiltrated information in operations targeting Zimbra Collaboration Suite users, particularly in campaigns exploiting CVE-2025-66376. Flowerbed functions as attacker-controlled receiving infrastructure rather than an endpoint payload, and is designed to ingest, log, and temporarily store stolen victim data delivered over both DNS and HTTPS.
The framework includes multiple coordinated components: Catcher, which acts as the primary HTTP and DNS collection service; Nginx, which fronts the infrastructure as an HTTPS reverse proxy; Certbot, which automates TLS certificate provisioning; and Gardener, which performs health-check and service-monitoring functions. Reported operations used Flowerbed to receive compressed mailbox data, account information, Global Address List contents, credentials, two-factor recovery material, and application passcodes stolen by the related Ulej exploitation capability.
Flowerbed has been associated with cyber-espionage activity against organizations in government, defense, education, energy, law enforcement, media, non-governmental, and technology sectors, with a focus on covert email intelligence collection. In the documented Zimbra campaign, exfiltration to Flowerbed occurred immediately after exploitation and used covert dual channels: smaller data elements were encoded for DNS-based transfer, while larger archives and structured data were uploaded over HTTPS. The infrastructure was operated on attacker-controlled virtual private servers and rotated regularly to reduce exposure and complicate tracking.
Associated vulnerabilities
1 CVE Mallory has correlated with this family across public research and vendor advisories. Each card links to the full Mallory page for that vulnerability.
Associated threat actors
1 distinct threat actor Mallory has associated with this family across public reporting. Open in Mallory to see the full evidence chain and overlapping campaigns.
Flowerbed est le framework de collecte côté serveur, un projet Python conteneurisé via Docker comprenant : Catcher : serveur DNS/HTTP recevant les données exfiltrées ; Certbot : génération automatique de certificats Let’s Encrypt ; Nginx : reverse proxy HTTPS ; Gardener : health check du service Catcher.
Techniques & procedures
24 distinct techniques documented for this family, organized by ATT&CK tactic.
Resource Development
6 techniques
Resource Development
T1583 — Acquire Infrastructure (Resource Development)
LAUNDRY BEAR utilise des VPS provisionnés via des identités fictives
Using a custom-developed capability [T1587.001] named “Улей” or “Ulej” (Russian for beehive), LAUNDRY BEAR successfully targeted and exfiltrated sensitive user information.
T1588.002 — Obtain Capabilities: Tool (Resource Development)
The simplistic Flowerbed codebase has indications that artificial intelligence (AI) played a role in its development ... [T1588.007].
T1608 — Stage Capabilities (Resource Development)
Initial Access
2 techniques
Initial Access
CVE-2025-66376 does not provide direct server compromise – its impact is authenticated-session abuse leading to mailbox theft, credential exposure, and account persistence.
the phishing bears abused the security hole in the Zimbra Collaboration Suite by sending malware-laden HTML email messages to target Western organizations
Persistence
1 technique
Persistence
CVE-2025-66376 does not provide direct server compromise – its impact is authenticated-session abuse leading to mailbox theft, credential exposure, and account persistence.
Privilege Escalation
1 technique
Privilege Escalation
CVE-2025-66376 does not provide direct server compromise – its impact is authenticated-session abuse leading to mailbox theft, credential exposure, and account persistence.
Stealth
1 technique
Stealth
CVE-2025-66376 does not provide direct server compromise – its impact is authenticated-session abuse leading to mailbox theft, credential exposure, and account persistence.
Credential Access
2 techniques
Credential Access
The attack doesn’t require any user interaction other than viewing the malicious email, and once that happens, the attackers get to work exfiltrating a ton of data. This includes the victims’ last 90 days of email communications, email addresses and passwords, the organizations’ email directories such as global address lists, two-factor authentication tokens, and newly created application passcodes.
Evidence details the creation of application passcodes, including passcodes named ZimbraWeb . If an attacker obtains credentials, active session material, application passcodes, or 2FA recovery codes, applying the software patch alone may not remove access that was already established.
Collection
6 techniques
Collection
The observed campaign reportedly targeted recent email, address-book information, account and mailbox metadata, 2FA recovery material, and application passcode functions.
Flowerbed est le framework de collecte côté serveur
T1114 — Email Collection (Collection)
T1114.002 — Email Collection: Remote Email Collection (Collection)
This includes the victims’ last 90 days of email communications, email addresses and passwords, the organizations’ email directories such as global address lists
T1560 — Archive Collected Data (Collection)
Command and Control
3 techniques
Command and Control
Smaller data is encoded and transmitted in DNS A-record queries, while larger payloads, including mailbox data, are uploaded over HTTPS as compressed archives to the attacker-controlled servers.
For smaller data items, the malware encodes information in Base32 and triggers DNS lookups through specially crafted image URLs.
The group... swaps out its infrastructure every 7 to 60 days, which makes long-term tracking harder.
Exfiltration
5 techniques
Exfiltration
once that happens, the attackers get to work exfiltrating a ton of data
L’exfiltration utilise deux canaux : DNS ... et HTTPS
HTTPS (JSON et octet-stream)
DNS (Base32, RFC 1035 compliant)
Stolen data lands on a backend the advisory calls Flowerbed, a set of Docker containers running on rented cloud servers. One container, “Catcher,” acts as both a DNS and HTTP server to receive the stolen data.
Observables
51 indicators attributed across vendor reports, sandbox runs, and researcher write-ups. Values are masked here and available in full in Mallory.
Recent activity
5 sources tracked across advisories, community write-ups, and news. New activity surfaces here as Mallory finds it.